If you’re interested in making your home heating greener and are considering switching to a heat pump, one of the first questions you might ask is how much it will cost to install. The answer to that isn’t clear, especially for people who have little prior knowledge. You might search online and find that a heat pump unit on its own costs about £5,000. On the other hand, you might come across newspaper headlines that suggest it could cost up to £35,000 for a heat pump installation. That’s a big window, and doesn’t tell us much.
We know from our earlier research into heat pump adoption that consumers really want personalised information that helps them understand the technology and how to make decisions. Without it, consumers are put off. Heat pump adoption becomes too complicated and people stick with what they know, which is usually fossil fuels.
Our aim is not to replace the job of professional heat loss surveys and installer quotes, rather it is to see if it’s possible to provide householders with a figure that is within a range which is accurate enough to help them better understand what’s possible and identify what to do next. Providing such information would, we think, build confidence and encourage more householders to adopt heat pumps.
At the moment, our aim is simply to test the concept and find out if providing such information is possible. There are various parts to this project, and we’ve assembled a mixed methods team including data science, design and analytical expertise to work on it.
We’ll be exploring the core cost components of a heat pump’s installation which we will find by speaking with heat pump installers. We’ll ask them about what they look for when they quote for jobs, and about the elements of the final bill. We’re keen to learn whether there are particular things that strongly influence the total cost and, if there are, what they are. Knowing this will help us understand cost estimations and produce our own, and we’ll approach this in two different ways.
One way will be through the creation of an arithmetical model based on user-collected data from householders themselves. Using the price influences that we identify from our interviews with installers, we want to test whether simply collecting a few key pieces of information about a property can predict heat pump installation cost with sufficient accuracy.
Another part of this project will be to create a machine-learning predictive model. This will draw on data from existing installations from Energy Performance Certificates (that all homes have) as well as registrations of heat pump installations on the Microgeneration Certification Scheme (MCS) database. We could use these to predict how much an installation might cost in a similar property. The model could be enhanced by the findings from the heat pump installers, for instance by guiding the model to pay particular attention to some characteristics of a property over others.
We’ll also experiment with the possibility of combining these two elements to see whether mixing user-entered data with the outcomes from the predictive model leads to improved accuracy.
It is by no means certain that it’ll be possible to generate the cost estimations that we want. We know that heat pump installations are complex and there are a many factors that affect cost. The fabric of older homes, as well as their heating systems, are the outcome of decisions taken by plumbers and builders since the day they were first built. It might be that there isn’t a way to produce useful predictions. This is a three-month experimental project, and we’d regard failure to achieve our aims as a useful learning opportunity.
If it is possible to create some “good enough” predictions, the next step will be to work on what users need. As well as knowing what the most significant factors are that affect cost, we’ll also need to know how possible it is for householders to collect information on those factors. We’ll want to see what can be collected through automation and what householders would prefer to find out for themselves. We’ll also need to test whether householders really find such information as useful as we think they will, and what the best way might be to deliver it to them.
We could also explore other uses for this study. If successful, the knowledge we acquire could be used to predict the costs of full decarbonisation of the UK’s heating infrastructure. It could help identify households where the take-up costs would be the lowest and then target those households with heat pump adoption guidance.
We’ll be working in the open throughout this project, updating with our findings as we get them.