Learnings

Outcomes

The impact of the pandemic on domestic electricity consumption:

-        We found three types of customers. One group did not change consumption at any point during the pandemic and carried on with pre-pandemic consumption patterns. This was approximately a third of the sample. The second group changed consumption during the pandemic but reverted to pre-pandemic consumption by the end of the period considered. This was approximately a quarter of the sample. The remaining customers changed consumption during the pandemic and did not fully revert to the pre-pandemic. The majority of these customers persistently increased daytime consumption, with a minority of this sub-group of customers persistently increasing peak consumption.

-        We found the overall demand and consumption increases to have a relatively minor effect on bills and were not necessarily widespread across customers. Therefore, the impact on fuel poverty was small and dwarfed by the impact of the subsequent cost of living crisis. We found that there was not a widespread impact on the network. The majority of areas would not be pushed into constraint as a result of the pandemic effect. However, there were a few areas where this might be the case (areas where the networks were already close to capacity, and there are significant proportions of people working from home). NGED could investigate these areas further to determine if its forecasting assumptions (e.g. within DFES) should be revisited.

-        Full report of this stream of work is available here: www.nationalgrid.co.uk/downloads-view-reciteme/627612

The impact of the cost-of-living crisis: 

-        We found that, similar to the impact of the pandemic, there were significant changes in consumption among customers. There was a large reduction in gas and electricity consumption during the three-month period in our study. The reduction in consumption was larger and more variable in gas (14%) compared to electricity (8%), but it was still a large change for both.

-        We found that the impact of increasing bills meant that the proportion of our sample of fuel poverty increased from 12.5% to 15.7%. This increase could have been even higher if it weren’t for households taking actions to reduce consumption. This has two implications. First is that there are likely to be many households on the edge of fuel poverty that may not receive support as a result of actions they are taking themselves to ensure they do not fall into fuel poverty. Second, it shows that actions can be effective in saving on household bills, but it might come at a cost of heating the home.

-        The aggregate changes in consumption mask the fact that some households did a lot to change their consumption and some did very little. For electricity, around a quarter of households reduce consumption by more than 20% while there are also a similar proportion of households who did not change their behaviour, consuming around the levels that would have been predicted. The variation in response is a clear signal that different households have different levels of energy price sensitivity.

-        While almost all household groups saw an increase in the number of households in fuel poverty, households with elderly people saw the biggest increase as over 35% more households fell into fuel poverty. This reflects the fact that households with elderly people made smaller reductions in consumption than average. Extra provisions may be required for these households.

Vulnerability prediction model:

-        A website was developed to demonstrate how the vulnerability detection model could be used in practice: https://venice.fnc.digital . The dashboard presents the smart meter data from ten real homes. The user can investigate the usage of each household and compare them between the houses. This allows the user to investigate the natural variations between time-of-day and type of household.

-        The results of the research undertaken by behavioural scientists to determine how customers in vulnerable situations may interact with their electricity differently is published here: www.nationalgrid.co.uk/downloads-view-reciteme/582871. The report covers the full methodology, findings, and conclusions of the behavioural science elements of the project, which were used to support data scientists in developing a model to identify vulnerability. 

-        Findings from the development of three separate models to identify if a household has a range of different vulnerabilities have been produced and available here: www.nationalgrid.co.uk/downloads-view-reciteme/608070 

Net-Zero carbon future scenarios for community:

-        Methodology for communities developing local net zero scenarios was produced and is available here: www.nationalgrid.co.uk/downloads-view-reciteme/638203 . The methodology can be replicated in whole, but there is also lots of potential for communities to use the document to consider the potential interactions between energy and vulnerability. 

-        Community scale net-zero community carbon accounting method (available here: www.nationalgrid.co.uk/downloads-view-reciteme/603289) and accompanying carbon tools are developed to help communities explore pathways to reach net zero futures. The tool is available here: www.nzcom.cee-uk.com 

-        A guide has been written to assist community groups interested in planning for net-zero to help the community to make the transition to a low carbon future and make sure that the most vulnerable members of the community are supported during this process. The guide is available here: www.nationalgrid.co.uk/downloads-view-reciteme/624101 

Community led business model:

-        A report summarising a community led business propositions and community owned initiative that could be used by any local community is available here: www.nationalgrid.co.uk/downloads-view-reciteme/638202 .

A strategy to reduce consumer energy costs with an innovative heat pump tariff and peak avoidance strategies was considered and available here: www.wren.uk.com/images/documents/NZCom/NZCom_WP7_Final_Report_v5.pdf

Lessons Learnt

Smart Meter Data 

-        Access to smart meter data was an ongoing challenge affecting both the impact of the COVID-19 pandemic work performed by Frontier Economics and modelling activities completed by Frazier Nash Consultancy. However, there are benefits to DNOs in accessing individual smart meter data which could provide benefits to customers.   

-        Having access to smart meter data meant using some analytical techniques that DNOs had not previously performed (i.e. linear regression and machine learning) to predict energy consumption. There are lots of potential applications for networks for using predictive modelling (i.e. establishing a baseline for assessing measures designed to promote energy efficiency or flexibility; or predict consumption level in response to extreme weather conditions).

-        Clustering analysis performed by Frontier Economics allows for grouping customers who do not necessarily share demographic characteristics but had similar consumption patterns which could help DNOs identify different ‘archetypes’ of consumption patterns, and then consider how the consumption of these groups may change in the future. This is particularly important in response to increased electrification in the home and with warmer summers having an impact on electricity consumption with air conditioning use. This type of clustering could also assist with anonymising smart meter data. 

Vulnerability prediction model

-        Models to detect households in vulnerable situations need to be validated using smart meter data from households with known vulnerability. It was not possible to perform validation during the course of the project due to data protection. This meant it was not possible to gain a detailed understanding of the underlying changes in behaviour that result in differences in energy consumption which is necessary to distinguish changes relating to becoming vulnerable from other changes that can take place in a household e.g. a change in the number of occupants, an occupant changing jobs etc. 

-        The use of a probabilistic model of smart meter data for the purposes of predicting vulnerability was supported, given the complex and transient nature of vulnerability. However, more granular smart meter data is required to identify vulnerabilities reliant on specialists’ appliances. 

-        Data gathered from smart meters should be considered in conjunction with other sources of available information, such as account history and known household characteristics. Triangulation is likely to improve the quality of predictions relating to vulnerability, and crucially, smart meter data seems likely to be capable of providing incremental utility to the accuracy of that prediction.

-        The impact of vulnerability is likely to vary greatly across households and time, and certain vulnerabilities may be more detectable than others.

-        The results of the cost-benefit modelling indicate that an accurate, reliable vulnerability identification model is very likely to provide a positive Net Present Social Value (NPSV). However, the extent of this will depend on the choice of algorithms and decisions on how to prioritise NPSV or benefit-cost-ratio, as well as the incentives provided to increase Priority Service Register spending.

Net-Zero carbon future scenarios for the community 

-        Multiple plausible energy system futures exist, as does the possibility of not reaching net-zero at all. By using the NG FES and DFES, we sought to use publicly available, robust and frequently updated scenarios that enabled replicability and cross-comparison at different levels of scale. However, feedback received during the production of scenarios revealed limitations and inappropriate underlying assumptions of these approaches to the community in question (e.g. assumptions around carbon sequestration, estimates of renewable updates, behavioural change and demand reduction).

-        Change is systemic, and which pathway is realised will be influenced by a huge number of interconnected technological, economic, behavioural and political factors. These factors can align to support decarbonisation, but they can also align to frustrate decarbonisation. Failing to upgrade buildings will mean net-zero targets are missed. It will be possible to address decarbonisation – to an extent - without addressing vulnerabilities. However, progress towards decarbonisation will be frustrated if we fail to obtain a broad social mandate for net zero, and/or existing vulnerabilities are neglected.

Community led business model

-        From the assessment work around community business cases, a support gap was identified with regards to vulnerable customers being able to access trustworthy information and grants that could help them to improve the energy efficiency of their home and take up low carbon technologies.  Typically for most of these schemes once the household has been deemed as eligible, they then need to obtain an up to date Energy Performance Certificate (EPC) costing upwards of £60, plus they (or their landlord) may also need to make a contribution to the cost of installation.  For some, this is unaffordable and therefore are unable to access this much needed support.

-        Pseudo microgrid: The advantage gained from a pseudo microgrid is strongly linked to the capacity and type of renewable energy connected. More local generation results in more advantage to the community.

-        With sufficient renewable energy generation on a pseudo microgrid, a significant energy cost saving is achieved in the ‘do nothing’ action group. This action might represent the fuel poor within the community over the next 10 years or more.