Smart Meter Enabled Thermal Efficiency Ratings (SMETER)


City Science was successful in obtaining funding from the Department for Business, Energy & Industrial Strategy (BEIS) to develop Smart Meter Enabled Thermal Efficiency Ratings (SMETER) as part of a consortium led by Hoare Lea. SMETER provides the first accurate, scalable, and continuous method of assessing and monitoring the thermal performance of the building stock and is a key component for addressing the decarbonisation of buildings.



Energy provision for space heating in buildings is a major source of carbon emissions in the UK whilst it can also detrimentally impact socio-economic outcomes such as fuel poverty. For example, 60-70% of energy use in a typical building in the UK can be attributed to space heating.

Typically, energy requirements for space heating in a particular building are governed by the efficiency of the heating system and the rate of heat loss. The size of a heated space and average insultation level are properties of buildings and are typically consolidated as part of a measurement called the ‘Heat Transfer Coefficient’ (HTC). For a given internal temperature and outside temperature, the HTC determines how fast heat leaks out of a building and hence is a critical parameter to be able to measure to assess the current thermal performance of the UK building stock. Accurately measuring the HTC is of paramount importance, not just for accessing a selecting fabric retrofit measure, but also for accurately sizing heat pumps. There are two common methods of measuring the HTC, however, both have opposing problems:

  • Co-heating test: Provides an actual measure of the HTC, however, it requires vacating the building for 3 weeks in the winter and therefore not practically scalable
  • Building Surveys: Provide a more scalable method of measuring the HTC, as the building can be in use, however, tend to be quite inaccurate due to overreliance on assumed values, including variations between building components and human error.

We developed SMETER to address all these problems allowing for the first scalable, accurate and continuous method of measuring the HTC. The unique challenge of this project was to enable the thermal performance of a building (Heat Transfer Coefficient) to be inferred accurately whilst the building was in use, in a way which is easily scalable and is not invasive for the building occupants like the current process. SMETER utilises Smart Meter data to measure the thermal performance (Heat Transfer Coefficient) of a building and monitors heat loss whilst the building is in use.

Our Approach

In phase 1 of the project, our energy and data science specialists developed state of the art core algorithms and software library for measuring the HTC via in-house and climate data streams. This uses data from smart meters to develop a simple report which can be used homeowners to demonstrate their home’s thermal performance. As part of the algorithm suite, we also developed automated methods to separate out the building average insulation level and building size components which both contribute to the HTC. Generally SMETER allows:

  • Energy inefficient buildings to be identified
  • The result of retrofit outcomes to be monitored Applied to a building stock SMETERs allow for:
  • An accurate highly evidenced based understanding of the energy efficiency state of the housing stock
  • An identification of the buildings most in need of retrofit measures in the housing stock
  • Continual monitoring of the housing stock as retrofit improvements are applied and the impact of those retrofits on energy and CO2 reduction

We tested SMETER through testing our solution on over 100 virtual dwellings, 30 real dwellings and through a wider market research process. Through this, we worked with Purremetrix who provided the hardware installation for data collection. We received excellent results and highly positive feedback.

Unique Challenges and Features

A unique challenge of this project was to enable the thermal performance of a building (Heat Transfer Coefficient) to be inferred accurately whilst the building was in use, in a way which is easily scalable and isnot invasive for the building occupants.

  • Data collection from buildings which are in-use have complex signatures which need to be broken down andanalysed to enable accuracy of the HTC estimate
  • Enabling scalability means that analysis and data collection must be automated
  • To be minimally invasive data needed to be extracted from minimal sensor coverage in the building and preferably from sensors that may already be reasonably integrated in the building (i.e. smart meters and temperature sensors)


We overcame the project challenges by combining building physics know-how with complex data science techniques to develop a cutting-edge algorithm suite to accurately predict the thermal performance of in use buildings. Importantly this measure is continually updated and automatically adjusts when fabric improvements are made to the dwelling. Minimal hardware was deployed in the households making the experience non-intrusive for residences. The data pipelines from the deployed sensors to the algorithm suite were fully automated allowing for a scalable solution.

heat imagery of a residential street