UCL Energy Institute


Energy use in the UK building stock: new empirically based models

Partners include: the Department of Energy and Climate Change, the Energy Saving Trust, the Ministry of Defence and the Chartered Institution of Building Services Engineers.

National plans for CO2 reduction and security of energy supply depend on very significant and rapid reductions in the UK building sector. These plans will need much better knowledge of present patterns of energy use in the building stock, and the incorporation of this understanding into new predictive models.

This project sought to contribute to developing this knowledge for the national stocks of both domestic and non-domestic buildings.

The existing CaRB database and model of the non-domestic stock at the level of individual premises, developed at UCL, was elaborated and strengthened with the incorporation of new data to create the CaRB2 model.

The initial intention was to build a separate new model with which to follow trends in energy consumption over recent years and to try to determine the various effects of weather, economic activity, growth in floor area, changes in fuel price, and efficiency improvements.

This goal was not however achieved, largely because of data problems and incompatibilities between the three stock models produced to date: the BRE’s N-DEEM and the two successive versions of the CaRB model.

These previous models operate just with floor areas and rates of energy use per unit of floor area. They do not deal with buildings as units, even though the geometry and construction of buildings are important for energy use. The project explored new methods for relating non-domestic floor areas to buildings and their construction, using information from digital maps, Valuation Office Agency rating data, 3D digital models of cities, and other databases. Much greater progress was made here than was initially envisaged, and a test model has been built for the London Borough of Camden. In this model, premises are assembled automatically into 3D representations of buildings, from which geometrical properties can be calculated, and data can be attached on other characteristics such as age and materials. The process is capable of extension to the whole country.

On the domestic stock: the team had previously carried out extensive analyses of current patterns of domestic energy use. There can be significant differences between the levels of energy savings predicted by theoretical models, and actual savings as observed from empirical measurements. There are likely to be several causes, including so-called 'rebound' effects, where the occupants react to energy improvements by for example enjoying higher temperatures or heating more rooms. Such behavioural effects can be observed through analysis of data derived from 'smart meters'. The project proposed to compare data for the same dwellings from smart meters with data from normal 'dumb' meters, in order to try to better understand these feedback phenomena. These effects could then be allowed for in improved predictive models. In the event there was a shift away from this goal due to a lack of comparable samples. Instead, work was undertaken on smart meter data using data from a field trial of approximately 80 dwellings that had received external wall insulation retrofits.

In research focused on using the Homes Energy Efficiency Database (HEED) containing data on approximately 13 million dwellings, an innovative approach based on medical epidemiology was used to structure the study designs and data analysis. The application of an epidemiological approach provided a way of developing empirical population-based evidence that was of relevance to the two remaining domestic project objectives: retrofit market segmentation, and empirical models of changes in demand due to retrofits.


Non-domestic building stock. CaRB2, a new bottom-up model of energy use in the non-domestic stock of England and Wales, was constructed with data for 2010. This is an updated and extended version of the previous Carbon Reduction in Buildings (CaRB) model built at UCL with data for 2004. Floor space data for CaRB2 are drawn mostly from the Valuation Office Agency’s Rating List and SMV databases, with additional information coming from many other commercial and public sources. Energy data are taken from surveys made by Sheffield Hallam University in the 1990s, and from Display Energy Certificates. Phil Jones of Building Energy Solutions collected further data for the model from numerous organisations. Unfortunately however data promised by the Ministry of Defence on their estate were not forthcoming. CaRB2 has not been published as yet, because of remaining uncertainties about some sub-sectors. But a preliminary version has been used by DECC and others in policy formulation and survey work, as described below.

An analysis was made for the Chartered Institution of Building Services Engineers of all Display Energy Certificates deposited between 2008 and 2012, and the results employed in building the CaRB2 model.

Much of the work in the project has been concentrated on developing a second, new type of bottom-up model, in which building geometry and construction are explicitly represented. The model’s structure depends on linking together building footprints in OS digital maps with entries for non-domestic premises in the VOA’s SMV database, by their addresses. This process is complicated by the fact that not all footprints are addressed, and have to be ‘captured’ by other means. However OS/VOA matching rates in excess of 90% have been achieved to date. The SMV database gives the floor area of premises by floor levels, and further breaks each floor down into sub-activities. This means that once address matches have been made, the floor space can be piled up on the footprints into ‘buildings’. Thus separate premises, for example those belonging to tenants in a multi-occupant office block, can be put together floor-by-floor into these buildings. Building heights are obtained from laser measurements. Geometrical properties such as volumes, exposed surface areas and plan depths can be taken off.

The method has been trialled in a case study of the London Borough of Camden. Data on materials, structural systems and building age, collected specially by the GeoInformation Group, are being attached to the buildings. A model of electricity use by equipment in spaces devoted to different sub-activities, developed previously for the city of Leicester, has been applied to Camden. A simulation model of heating and cooling demand based on EnergyPlus software is under development. DECC has made available gas and electricity meter data at the premise level against which these models can be tested and calibrated. The overall approach represents a methodological breakthrough in our ability to use existing datasets to describe and model the entire building stock of the UK, and means that an ‘epidemiological’ approach to energy use in non-domestic buildings becomes possible for the first time. Since all the methods for constructing and analysing the Camden model are automated – or capable of being automated – the process could in principle be rolled out to the whole of England and Wales. This work has clearly demonstrated the feasibility of creating bottom-up national building stock models based on geometrical, age, activity and energy data for all buildings. We expect the progress that has been made to be of great value in advancing our understanding of the drivers of energy use in this relatively poorly understood sector.

Domestic building stock. In the domestic research, the project sought to address three objectives: examining high-frequency meter data (smart meter data); examining changes in energy demand due to energy efficiency retrofits and comparing actual to modelling changes; and examining the composition of the retrofit market.

The research on high-frequency energy data studied the impact of retrofits on energy demand using smart meter data from a field trial of 80 dwellings. The analysis examined the relationships between external environmental conditions and occupant practices through the data. The development of an energy demand response curve provided a method of understanding the relationship between dwelling energy performance and occupant practices. Changes occurred in the response curve following an energy efficiency retrofit, which might not otherwise be seen in annualised demand. The demand curve concept, as implemented at UCL, is a significant step forward: theoretically, because it is independent of time scale; and practically, because it relates directly to the question of peak loads on national energy systems – an issue which we expect to become more important in coming decades.

An epidemiological approach (i.e. population-based research designs and analysis) was applied to the large HEED data and annualised energy demand data. This approach was selected as a way of dealing with both a large dataset and also the ability to examine differences in energy demand and changes in energy following retrofits between sub-groups. The research examined factors that affected both energy demand and energy savings following retrofits, and identified physical and household factors that would be expected to affect savings. This provided a method of identifying factors that were important for households that did and did not achieve expected modelled savings. The research has important implications for government programmes that mix actual energy savings and expected modelled savings. The epidemiological approach was also applied to the market segmentation work, which used an ecological study design to examine the relationship between the uptake of energy efficiency retrofits and neighbourhood level characteristics. The research showed that high rates of uptake were dependent on the programme through which they were delivered, and also on household factors such as ownership and income levels.

Publications. Members of the project team have published nine journal and conference papers, and two more are in preparation. They have produced two reports and a special issue of Building Research and Information in honour of Harry Bruhns.

  • Ekins P, Dresner S, Browne J, Preston I, White V, Hamilton I G, (2012) Designing Carbon Taxation to Protect Low-income Households, London, UK: Joseph Rowntree Foundation; 124 pp.
  • Evans S, Liddiard R and Steadman P, ‘A 3D geometrical model of the non-domestic building stock of England and Wales’, paper submitted to Building Simulation and Optimization 2014 Conference, London
  • Hamilton, I G, Shipworth, D, Summerfield, A J, Steadman, P, Oreszczyn, T, and Lowe, R J (2014), ‘Uptake of energy efficiency interventions in English houses, 2000 to 2007’, Building Research and Information, Online. doi:10.1080/09613218.2014.867643
  • Hamilton, I G, Summerfield, A J, Lowe, R J, Ruyssevelt, P, Elwell, C, and Oreszczyn, T, (2013), ‘Energy epidemiology: a new approach to end-use energy demand research’, Building Research and Information, 41(4), 482–497. doi:10.1080/09613218.2013.798142
  • Hamilton, I G, Steadman, P, Bruhns, H, Summerfield, A J, and Lowe, R, (2013), ‘Energy efficiency in the British housing stock: Energy demand and the Homes Energy Efficiency Database’, Energy Policy, 60, 462–480. doi:10.1016/j.enpol.2013.04.004
  • Hamilton I G, Summerfield A J et al. ‘The impact of solid wall insulation on energy demand’, Energy and Buildings (in preparation).
  • Hamilton I G, et al. ‘Housing energy savings and energy efficiency in England: a population-based retrospective case-control study’, Building Research and Information (in preparation)
  • Hong, S-M and Steadman P (2013), An Analysis of Display Energy Certificates for Public Buildings, 2008 to 2012, Report to the Chartered Institution of Building Services Engineers, October, 73 pp.; available at http://www.bartlett.ucl.ac.uk/energy/news/documents/CIBSE__Analysis_of_Display_Energy_Certificates_for_Public_Buildings_.pdf
  • Hong S-M, Paterson G, Mumovic D and Steadman P (2014), ‘Improving benchmarking comparability for energy consumption in schools’, Building Research and Information 42, Jan/Feb, pp.47-61
  • Isaacs N and Steadman P, eds. (2014) special issue of Building Research and Information, ‘Understanding energy and the non-domestic building stock’ in memory of Harry Bruhns, Vol.42 January/ February
  • Jensen, H T, Keogh-Brown, M R, Smith, R D, Chalabi, Z, Dangour, A D, Davies, M, … Haines, A, (2013), ‘The importance of health co-benefits in macroeconomic assessments of UK Greenhouse Gas emission reduction strategies’, Climatic Change, doi:10.1007/s10584-013-0881-6
  • Laurent, M H, Allibe, B, Tigchelaar, T, Oreszczyn, T, Hamilton, I. G, and Galvin, R, (2013), ‘Back to reality: How domestic energy efficiency policies in four European countries can be improved by using empirical data instead of normative calculations’, in ECEEE Summer Study: Rethink, Renew, Restart.
  • Liddiard R (2104), ‘Room-scale profiles of space and use and electricity consumption in non-domestic buildings’, Building Research and Information 42, Jan/Feb, pp.72-94
  • Steadman P, Hamilton I and Evans S (2014), ‘Energy and urban built form: an empirical and statistical approach’, Building Research and Information 42, Jan/Feb, pp.17-31


The preliminary version of the CaRB2 model, and analyses of VOA data, have been supplied to DECC and their consultants Verco and GfK, for use in the planning and conduct of DECC’s current Building Energy Efficiency Survey (BEES) programme to collect new data on the non-domestic stock. Classifications of activities developed by the UCL team are being used in this project.

Advice on the non-domestic stock, and analyses of data sets, have also been supplied to a wide variety of public and private sector organisations. Display Energy Certificate data have been analysed for CIBSE as mentioned.


Details of grant on EPSRC website (opens new page)

carbon modelling