We gathered data about residential properties in Great Britain to map out how suitable different kinds of heat pumps are at the neighbourhood scale. This will help Nesta and other organisations in the energy sector identify the best approach to decarbonising heating in each neighbourhood. In turn, this will also support our switching streets approach to delivering low-carbon heat.
Our aim was to create a robust, impartial and reliable dataset showing how suitable different types of heat pump are for each neighbourhood. This dataset could be used to challenge common assumptions about heat pump suitability and understand which locations are suitable to make the switch.
This aim was not to replace the need for local planning for heat, but to simplify how strategic choices are made about which low-carbon heating technologies are suited to different neighbourhoods.
Understanding which homes are suitable for which heating technologies is important for policy makers, local authorities and homeowners. Consumer choice, market offerings and infrastructure development will all be guided to some extent by data and models that present assumptions about what works where. These assumptions are largely choices made by whoever owns or builds the model and are rarely impartial – so without unbiased evidence to challenge this they tend to be accepted. For example, air source heat pumps are ruled out in homes where they could be installed, or they are assumed to be viable in homes where there may be other options. We want to facilitate a debate about the best technology choices for different types of homes.
Having reliable data on how suitable areas are for different low-carbon heating technologies is a critical tool to switch households away from fossil-fuel heating. This project worked directly in combination with our work on clean heat neighbourhoods, which is developing an alternative model for rolling out low-carbon heating. However, a street-by-street approach is only effective with good data to identify which streets to target.
In previous phases of this project we used data science approaches to modify and enrich the openly available energy performance certificate (EPC) dataset of properties. We undertook stakeholder workshops to understand the landscape of assumptions about heat pump suitability, and used these assumptions to map out suitability for different technologies at small-area levels.
Part of our work addressed the fact that EPC data is incomplete - it only includes around two thirds of residential properties in England and Wales. Therefore, we employed statistical techniques, such as iterative proportional fitting, to assign weights to the EPC dataset against more complete datasets from the 2021 UK census. The resulting dataset is more representative of all properties in an area, not just those captured in the EPC data.
A stakeholder workshop helped us obtain a clear view on the property and area-level factors affecting heat pump suitability and what should be measured to quantify those factors. Following this, the EPC dataset was enhanced to include these factors by linking EPC properties with other datasets.
After producing this enhanced EPC dataset we calculated heat pump suitability scores per property, and then by using the weights assigned to our dataset, we were able to create robust heat pump suitability average scores per neighbourhood. You can view the latest version of this dataset and a map to explore the suitability of different heat pump technologies across the UK (both the dataset and map won’t be updated in future, as we move towards building a tool for local clean heat planning). We also published key insights about suitability across the UK from our data in this blog and this deep dive into understanding how flats in England and Wales are heated.
In this phase of the project, we conducted user research and worked alongside Plymouth City Council to understand how our dataset and visualisation tool can be as useful as possible for local areas. The following changes will help us to hone our approach to mapping heat pump suitability:
Moving from having independent suitability scores for different heat pump technologies to identifying the most suitable technology for each area.
Our published suitability map allows users to select a specific heat pump technology and see how areas of the GB compare to each other in terms of suitability for that technology or which areas might be most suitable to it. However, results can’t be compared across technologies, meaning that suitability could not be compared across technologies. We’re now working on identifying the most suitable technology for each area, to make it more actionable for local areas. When we finish this work, the existing heat pump suitability map will be retired.
Identifying suitability at smaller neighborhood level, rather than for fixed lower layer super output areas (LSOAs)/ data zone level - which tend to have a minimum of 400-1,200 properties.
It is likely that different neighbourhoods within the same LSOA/ data zone have different low-carbon technology needs or needs which intersect with those of a neighbouring LSOA. Moving from grouping properties according to larger pre-existing LSOA boundaries to grouping properties in bespoke small areas provides flexibility and enables local areas to participate in this process. We will use a data science method called clustering to group similar homes at a small neighbourhood level. We’ll work with local authorities to manually label the most suitable technology for a group of clusters to help us evaluate and improve the models.
Conducting this work at a small neighbourhood level requires a complete record of all properties in a small area. EPC data is incomplete, while the OS open UPRN dataset contains every addressable location in the country. Using this dataset is associated with its own challenges, and it will require employing data science techniques to identify which subset of properties represent domestic properties, as well as imputing information about properties where unknown.
Adding the ability to visually verify data.
Early testing has shown that, in addition to mapping the most suitable technology for a group of properties and showing information about the properties, users want to be able to visually verify the reasons behind the assigned technology, by having access to satellite imagery and street view photographs.
Prototyping on one area before scaling up.
In the next few months we will use the learnings from our work alongside Plymouth City Council to focus on building a tool to better support local clean heat planning in Plymouth. We’ll also engage with other local authorities to test the underlying methodology and assumptions about the value, feasibility, and replicability of this in other areas, so that we can eventually scale the work to other areas of the UK.
To learn more you can read about the local clean heat planning tool we’re building.
Our work in the past few months has shown that identifying the most suitable technology for each home is complex and requires both data analysis and human expertise.
Technology mapping must incorporate an understanding of household needs and fuel poverty levels, and when undertaking energy planning and mapping suitability of low-carbon technology, it is key to have local context, including market mechanisms and demographics.
If you want to know more about the work we’re doing next, read the details about the tool we’re building - a tool to support local clean heat planning.
www.nesta.org.uk/project/mapping-heat-pump-suitability-across-great-britain/
