Lessons from smart tariffs: electricity consumption profiles

Introduction

At Centre for Net Zero, an autonomous research unit founded by Octopus Energy, we believe a “people-centred” lens is important for the future energy transition. In order to achieve a faster, fairer and more affordable transition, we must consider how electricity usage varies across different consumers, particularly in relation to the ownership of different low carbon technologies (LCTs), such as electric vehicles and heat pumps. By deeply understanding these customer behaviours, there are benefits of more accurately understanding their impact on the grid, potentially saving infrastructure costs by incentivising shifting demand through flexibility, and providing fairer prices for the consumers [1].

As more firm generation (such as gas and coal) is replaced by renewable variable generation (such as wind and solar) to achieve decarbonisation targets, it is essential to understand the requirements the grid has to face in times of high demand, commonly referred to as peaks [2]. These peaks usually occur when residential consumption is high and are related to activities such as cooking and heating. As we electrify energy-intensive sectors such as transport and heat [3], it is crucial that we understand what factors and behaviours cause these peaks in energy demand, how these peaks may change in the future and how new LCTs can be used as an opportunity to manage the peaks by either reducing it or shifting it to when there is abundant renewable energy, and thereby alleviate the stress on the grid [1].

 

Smart meter data is fundamental to understanding present and future energy consumption. Through Octopus Energy, we have access to smart meter readings from millions of customers from many parts of the world and, through surveys, we also know what LCTs some of these customers own. Smart meters enable us to understand a richer picture of how demand fluctuates with time, and “time-of-use” tariffs can help reward people for using energy at times when electricity is greener and cheaper. As customers adopt more LCTs, it is important to understand:

 

  • when the highest electricity usage occurs, i.e. when the peaks occur,
  • how the peaks change depending on the LCTs in a household, and
  • the impact that tariffs will have on shifting the peak.
 

In this analysis, we will investigate electricity consumption, also known as demand profiles, of UK households using smart meter data. We discuss how these demand profiles change over a day, over the year and with LCT ownership. We will also discuss the risks and opportunities they present to the energy transition.

Findings

Energy consumption is measured in kilowatt-hours (kWh), or watt-hours (Wh), in half-hour intervals in the UK electricity system. 1 kWh is equivalent to using 1kW of power for 1 hour. Since larger homes use more energy, we discuss energy consumption per square area (in m2). This enables us to compare consumption across different households even when they vary in size.

 

Customers use up to twice as much energy during the evening peak compared to the morning peak

Conventionally, customers with no LCTs use the highest amount of electricity during the evening peak,  roughly between 5-7pm. As most people do not own LCTs, the evening peak is also historically known as the grid peak. Before the pandemic, we would have also expected a smaller morning peak between 7-9am, but as more people work from home, the electricity consumption does not appear to drop after 9 am.

 

As shown in Figure 1, customers use between 1-4 Wh/m2 during evening peak, with a median usage of 2 Wh/m2. This is between 30 – 60% higher than the electricity consumption during the morning peak. Thus, a typical home of 90m2 would consume 0.75 kWh during the evening peak. A similar house would use 0.5 kWh during the morning peak.

Figure 1: A graph showing the median daily electricity consumption (Wh/m2) of households without any low carbon technologies. The band represents the 25th and 75th percentile of consumption. The vertical bands denote grid constraint times, in the morning and evening. The smart meter readings are from March 2021 to March 2022.

 

For customers with no LCTs, the variability in usage is also highest during the evening peak. This means that the amount of electricity used can change a lot between customers and between days. For customers who can use their electricity flexibly, it may be possible to incentivise them to shift some of their usage away from grid peak, though some energy intensive activities, such as cooking, may be harder to shift.


Evening peak usage is 56% higher in the winter compared to the summer

Figure 2 shows that part of the variability seen above can be explained by seasonality; in general, electricity usage is highest in winter, and lowest in summer. Median evening peak consumption is 2.5 Wh/m2 in winter, compared to 1.6 Wh/m2 in summer (56% higher). This is due to increased electrical heating demand in the winter. Thus, a typical house would use 0.7 kWh during the winter evening peak and 0.45 kWh during the summer evening peak.

Figure 2: A graph showing the median daily electricity consumption (Wh/m2) of households without any low carbon technologies. The coloured lines show how this median varies by season. The vertical bands denote grid constraint times, in the morning and evening. The smart meter readings are from March 2021 to March 2022.

Note: This graph is interactive: you can click the legend to isolate a specific season, or Shift+click to select multiple seasons. Clicking out of the legend resets the graph.

 

In both winter and autumn, grid peak is between 5-7 pm however in spring and summer the peak is smoothed out over several hours, likely due to households cooking later in the evenings when it is brighter.

 

Overall, the demand profile is the same across the seasons, with winter typically having a higher load during the day, mostly likely driven by customers who have electric heating.

 

Ownership of EVs increases daily consumption by 20%

As we transition to a more electrified future, customers are likely to have a more diverse mixture of LCTs. Through surveys of a subset of Octopus Energy customers, we have knowledge of the LCTs that customers own or use in their home. Combining this with their smart meter readings, we get a glimpse at what demand profiles might look like.

 

Figure 3: A graph showing the distribution of daily electricity consumption (Wh/m2) of households based on the low carbon technologies they own, sorted in ascending order of median daily consumption. The box represents the 75th and 25th percentiles, and the whiskers represent the 95th and 5th percentiles of consumption. The smart meter readings are from March 2021 to March 2022.

 

Figure 3 shows the daily variability in consumption, by LCT type. We see that customers with solar panels consume less than those without any LCTs, on average. This is because solar panels generate electricity for you, meaning you import less from the grid.

 

The median usage per day for customers with heat pumps is 100 Wh/m2, but varies from 50 – 175 Wh/m2. This translates to a median daily consumption of 168 kWh for a typical house but varies from 84 kW – 294 kW. This range is larger than for customers with no LCTs since some of these customers may not have electric heating, and thus have less variable electricity consumption. Compared to storage heaters or electric heating generally, heat pumps are an extremely efficient form of electric heating – they can convert 1 kWh of electricity into 3-4 kWh of “heating demand” [4], meaning your heating running costs will be between a third to a quarter of electric radiator running costs.

 

Customers with EVs and storage heaters consume more electricity per day. Customers with EVs use between 70 – 220 Wh/m2 with a median of 125 Wh/m2. Customers with storage heaters use between 95 – 310 Wh/m2 with a median of 175 Wh/m2. To put these figures into context, households that vary between 60 – 90m2 will use between 0.42 kWh – 19.8kWh if they own an EV, and between 5.7 kWh – 27.9kWh if they own storage heaters (assuming the lower size household uses the lower quartile of consumption, and the higher size household uses the upper quartile of consumption). The upper quartile of consumption typically reflects days on which customers charge their electric vehicle or are heating their home, respectively.

 

Customers with multiple LCTs use slightly less on average than those owning just an EV, likely because customers may have an EV and solar panels (the former increasing their consumption, but the latter reducing it).

 

There is greater variability in consumption among customers who own multiple LCTs, those who only own EVs and those who have storage heaters, than the other LCT categories considered. For those with storage heaters, this may be because storage heaters can vary a lot in age and efficiency. For both storage heaters and electric vehicles, customers do not heat their homes/charge their vehicle every day which explains some of the variability. The variability from multiple LCTs is due to the different combinations of potential technologies. 

 

This variability could be a challenge from a demand forecasting perspective, especially as more customers buy EVs, but it also presents an opportunity: if EV owners do not need to charge their cars every day, then the grid can utilise this to encourage flexible charging patterns by incentivising customers to delay charging if electricity demand is, or charge their EVs for shorter periods but on more days.

 

Solar panels can reduce weekly consumption by up to 50%

Patterns of consumption vary not only by season, but also by type of LCT. Figure 4 shows the total consumption per week in Wh/m2 of customers with EVs, customers with solar panels and customers with no LCTs.

 

Figure 4: A graph showing the distribution of annual electricity consumption (kWh/m2) of households split out by low carbon technology. The x axis shows the week number in the calendar year (e.g W05 is the 5th week of the year). The line shows the median total weekly consumption, with the band showing the 25th and 75th percentiles. The smart meter readings are from March 2021 to March 2022.

Note: This graph is interactive: you can click the legend to isolate a specific LCT, or Shift+click to select multiple LCTs. Clicking out of the legend resets the graph.

 

Prior to the pandemic, we would have expected more of a U-shape to the electricity demand of customers with no LCTs due to increased use of electricity in the winter months, for heating, lighting, etc. While this is still largely true, more people working from home may have led to the U-shape becoming a little more shallow.

 

Having solar panels reduces electricity consumption but to varying degrees across households. In summer, the median weekly usage is around 400 Wh/m2 for homes with solar panels compared to 600 Wh/m2 to those with no LCTs, a 30% reduction. For a typical house of size 90m2, this translates to 36 kWh if it had solar panels compared to 54 kWh if it had no LCTs. However, usage is roughly similar between November and March when solar generated electricity is lower. This may be because homes with solar panels are likely to have more electrified components such as electric heating and cooking.

 

Unsurprisingly, ownership of an EV increases the electricity consumption roughly 200  Wh/m2 per week throughout the year. This translates to 936 kWh for a typical house of size 90m2 over a full year (assuming consistent weekly charging patterns). While this increased demand can be a risk, particularly as the number of EVs increase, EV owners can be incentivised to charge their vehicles when demand for electricity is low through Time-of-Use tariffs.  We will take a look at this in more detail in a future deep dive.

 

Households with EVs can consume 10 times more electricity in a given period, but do so when demand is low if on an appropriate tariff

We can directly compare how the demand profiles change over the course of a day compared to the median consumption of households with no LCTs in Figure 5. This allows us to answer what would happen if customers got an EV, solar panel, or heat pump.

 

Figure 5: A graph showing the relative electricity consumption, compared to the median profile of households with no LCTs, split out by LCT category. The solid line shows the median relative daily consumption, with the band showing the 25th and 75th percentiles. The dashed line represents the relative median consumption of no LCTs. The smart meter readings are from March 2021 to March 2022.
Note: This graph is interactive: you can click the legend to isolate a specific LCT, or Shift+click to select multiple LCTs. Clicking out of the legend resets the graph.

Customers who have batteries and/or solar panels have median consumptions  4 – 6 times higher than those without any LCTs, however this consumption is confined outside of grid peak. This is because Octopus Energy offers smart time-of-use tariffs which are cheaper overnight, when overall demand is low. Moreover, these customers use less electricity throughout the day than those with no LCTs. While the total daily consumption is roughly comparable, this shows that customers who own batteries and solar panels can offer flexibility which can be used to help reduce or shift peaks, if they are on an appropriate time-of-use tariff.

In comparison, the median electricity consumption of those with EVs and heat pumps is only 1.5 times higher than the baseline overnight. Similar to battery owners, those with heat pumps can reduce their consumption during the day when demand may be higher.

 

While the median electricity consumption of EV owners is not much higher, the 75th percentile is 10 – 12 times higher than the median of those with no LCTs. This is not necessarily cause for concern because these are the customers who are most likely to have flexibility around their electricity usage. 

 

Potential for flexibility

 

Figure 5 shows that customers with EVs and batteries use the highest amount of electricity but they do so when demand is low. This is because their tariffs incentivise them to do so. 

 

For example, customers with EVs charge in the early hours of the morning because Octopus offers an EV charging tariff called Go that enables cheap charging between 00:30 – 04:30. There are also variants of this tariff called Go Faster, which offer different windows of different lengths. 

 

Some customers might also be on the Octopus Agile tariff, which has half-hour time varying prices that update each day and follow wholesale prices. Typically, electricity is cheaper early in the morning and more expensive around grid peak. This suggests that customers are able to shift their consumption to different times depending on the LCT they own. Through more dynamic pricing, selling surplus electricity back to the grid., etc, it is possible to spread electricity consumption more equally throughout the day while keeping prices fair for customers. This “demand shifting” is one aspect of flexibility.

 

Next steps

This analysis presents a first look at the types of demand profiles that emerge when customers are differentiated by technology type. This will become increasingly important in the future, both for grid planning and understanding how customers can be encouraged to use electricity flexibly.

 

In this research, we observed high levels of electricity demand from EV customers after midnight, as a result of their tariff. However, if every EV owner was on the same tariff, it would create a new grid peak. Octopus have therefore rolled out tariffs such as Go Faster, as well as a new beta tariff, Intelligent Octopus, which allows Octopus to control the charging levels while still meeting customer needs. This data set gives us a whole range of new insights into charging behaviour including when people plug in and how long it takes to charge. Using this it might be possible to harness the flexibility without creating a new grid peak. The results presented here are consistent with studies that suggest EVs are one of the largest sources of domestic flexibility [5]. 

 

In our next research project we will explore these EV charging behaviours, and start to paint a picture for how consumers with EVs, if on the appropriate tariffs, can use electricity flexibly saving them money, reducing CO2 and mitigating the grid impact by shifting the times that they charge. In particular, we will analyse the charging patterns and the customer behaviours under Octopus’ new beta tariff, Intelligent Octopus, which creates automated charging schedules for electric vehicles to reach a desired battery state of charge by a certain time at the least possible cost.

 

Key insights

  • For customers with no LCTs, electricity demand is currently highest during grid peak (5-7pm). These grid peaks may change as more LCTs become more prevalent in the future however, it is important to know when these are and how consumers can be incentivised to use electricity outside of peak times.
  • During grid peak, customers with no LCTs use 56% more electricity in winter than in summer but grid peak remains roughly between 5-7 pm across all seasons.
  • There is a 20% increase in the median daily electricity used by EV owners but the upper and lower quartiles increase by 50 – 60%.
  • However, this is outside of current grid peak hours as a result of their tariff. If left unmanaged, these consumers can create a new grid peak. When planning the future energy system, it will be important to design mechanisms that allow consumers to shift and spread the load conveniently and efficiently throughout the day.
  • The median electricity consumption in a half hour with a battery can be 6 times higher than if you don’t have any LCTs but these
    customers tend to use half as much electricity throughout the day offering a lot of flexibility to the grid.
 

About the data 

For this study, we reviewed aggregated smart meter data from 24,000 UK customers who are on a smart Time-of-Use tariff and analysed their electricity usage between March 2021 and March 2022. It should be noted that this included  periods of Covid lockdown during which guidance was in place encouraging people to work from home.  The result of this was greater energy consumption during the middle of the day for many households compared to the pre-pandemic period.  We also used publicly accessible EPC data to get the floor area of households to make comparison between households of different sizes possible.

 

Throughout the 2021, Octopus Energy surveyed their customers on smart tariffs and collected information on the types of LCTs that they own. We analysed this data in aggregate to understand the key trends and behaviours from each segment.

 

To focus this analysis, we only considered customers with no LCTs or one type. Since solar is often installed with a battery, we include customers with solar PV and a battery and label them as Solar PV customers. 

 

There are thousands of Octopus Energy customers who own two or more LCTs (for example an EV and solar panels), but since we’re presenting aggregated data, we wanted to showcase trends by isolating each type of LCT.  

 

The future energy system will need to be able to engage customers to use their electricity more flexibly. This is not just to help shift and spread demand but also to use renewable electricity when available and reduce dependence on coal and gas. It is our belief that by opening up data, more research can be conducted using this data, which will help make the future energy system fairer and greener. You can find the aggregated profiles and statistics used in this report here.

 

Interested? Get in touch

Interested in suggesting future deep dives or projects? Got your own ideas about flexibility and time-of-use tariffs that you want us to explore? We’d love to hear from you. Get in touch with us at info@centrefornetzero.org

 

References

[1] ‘Flexibility in Great Britain’, Carbon Trust. https://www.carbontrust.com/resources/flexibility-in-great-britain (accessed May. 12, 2022)

[2] D. P. Jenkins, S. Patidar, and S. A. Simpson, ‘Synthesising electrical demand profiles for UK dwellings’, Energy and Buildings, vol. 76, pp. 605–614, Jun. 2014, doi: 10.1016/j.enbuild.2014.03.012.

[3] ‘Energy consumption in the UK (ECUK) 1970 to 2020’, UK.GOV. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/1061644/Energy_Consumption_in_the_UK_2021.pdf (accessed May. 12, 2022) 

[4] Heat pumps produce heat by moving thermal energy from one fluid to another. To heat a building, this usually means moving heat from outside to inside. The Coefficient of Performance (COP) is a metric which estimates the amount of heat “produced” by a heat pump compared to the amount of electricity needed to produce that heat. An average COP of 3-4 means that it produces between 3-4 times more heat than a perfectly efficient conventional boiler for the same input electricity.

[5] ‘Crowdflex – Phase 1’, National Grid ESO. https://www.nationalgrideso.com/document/230236/download (accessed May. 12, 2022)