Bridging top-down evidence and bottom-up insights: Enhancing energy performance of Educational Buildings

Abstract

To achieve the transition to net-zero carbon buildings, there is an urgent need for cost-effective and energy-efficient solutions to improve the performance of school buildings, in response to which the Department for Education (DfE) has launched different school rebuilding and refurbishment programmes. The aim of this paper is to assess the energy performance of school buildings in the UK through top-down analysis of historical energy consumption data and bottom-up modelling of building energy consumption. Top-down analysis refers to understanding the recent energy use characteristics of primary and secondary school buildings constructed under the Building Schools for the Future (BSF) and Priority Schools Building Programme (PSBP) based on the Display Energy Certificate (DEC) dataset and Getting Information on Schools (GIAS). The bottom-up analysis is conducted by modelling one PSBP primary school to investigate the gap between the actual operational performance and the energy saving potential of this new school under the latest rebuilding programme. The study found that PSBP schools with compact and economical designs had higher overall energy performance, while BSF schools with higher capital investment operated more inefficiently. The results of the individual building simulations showed that even for the well-performing PSBP schools, there was still a significant performance gap between the actual energy used and the design expectations.

Introduction

The UK Department for Education (DfE) has launched three major school rebuilding programmes over the last 20 years, all of them aiming to construct or refurbish school buildings to improve building performance in line with the newly issued school design guidelines.Building Schools for the Future (BSF), a programme with a total budget of £55 billion, was launched in 2004 with the ambition to refurbish or rebuild all secondary schools in England over the next 15 years (Education and Skills Committee, 2007). However, this programme was stopped in 2010 for a number of reasons, including insufficient progress and cost overruns (DfE, 2012). Then the government allocated £4.4 billion to create the Priority School Building Programme (PSBP) in 2011 to deal quickly and cost-effectively with 537 schools in urgent need of repair (ESFA, 2016; James and Brown, 2013). The most recent is the School Rebuilding Programme (SRP) beginning in 2020, which aims to achieve net-zero carbon emissions from the operation of buildings, intends to rebuild and refurbish 500 schools and sixth form colleges in England to improve the sustainability of building performance (DfE, 2021). The government provided £2 billion in funding for the first 100 projects (DfE, 2021).Under the current climate challenges, RIBA advocates that the operational energy performance of school buildings should be less than 70 kWh/m²/year by 2025 and less than 60 kWh/m²/year by 2030 (RIBA, 2021).

Previous case studies have shown that the actual operational performance of school buildings generally did not match the design expectations (Bordass et al., 2001; Burman et al., 2018; Carbon Trust, 2011; Jain et al., 2020; Jradi, 2020; Pegg et al., 2007).It is therefore necessary to investigate and analyse the actual energy usage at the operational stage of buildings under various rebuilding programmes.The Energy Performance of Buildings Regulations (England and Wales) were first published in the UK in 2007 and introduced Display Energy Certificates (DECs) for the purpose of recording energy consumption and carbon emissions in public buildings (DCLG, 2007).Following subsequent amendments to the regulations, the need to display a DEC for public buildings over 250 m² in England and Wales has been a mandatory requirement since July 2015 (DCLG, 2012).The DEC database has now been used as a publicly available information in many studies to quantify the trends and characteristics of energy use for various types of public sector building stock in the UK (Armitage et al., 2015; Godoy-Shimizu et al., 2011; Hong et al., 2018, 2022, 2014).

This study intended to inform further improvements in the energy performance of the school building stock by analysing the characteristics of school energy use from both a top-down and bottom-up analysis. Specifically, this study seeks to address the following key questions:

·         What is the energy performance of school buildings under different rebuilding programmes in the UK?

·         Are the best performing schools in the top-down analysis fully realising the potential energy savings in actual use?

·         What performance gaps and improvement opportunities can bottom-up building modelling reveal?

Methodology

Top-down analysis approach

The top-down analysis involved using the DEC database and GIAS database to quantify the latest energy performance of primary and secondary school buildings in England. Three databases containing basic school information and DEC records were developed, which are for the All School Building Stock, the BSF School Building Stock and the PSBP School Building Stock. Python 3.11 was used to first preprocess the DEC dataset and then integrate with the GIAS dataset through address matching (DfE, 2024; DLUHC, 2024). The DEC dataset records annual energy consumption data for electricity and fossil thermal energy. The GIAS dataset provides the phase of education (e.g. primary or secondary). Table 1 presents the number of schools by school type for each year in the final unified dataset. Of these, other schools refer to school buildings other than primary and secondary schools, including nurseries, all-through schools and 16-plus schools. Due to inconsistencies in the submission dates of DEC record by schools, the latest subset of energy usage includes records from the unified database for the years 2022–2024. The database of all school building stock was first analysed to understand latest overall school energy performance. As the energy data did not conform to a normal distribution, a non-parametric Mann-Whitney U-test was used to assess differences in electricity and fossil thermal energy use between secondary and primary schools (Mann and Whitney, 1947). If the p-value is ≤ 0.001, it is considered that there is a significant difference between the two sets of data. The fossil thermal energy consumption data has been weather-corrected to 2021 heating degree-days, which represents the average climate for the UK (CIBSE, 2008). Next, the same statistical analyses were performed on the unified dataset of BSF and PSBP schools. Finally, the difference in energy consumption between modern schools under these two school rebuilding programmes was compared with the rest of the school building stock.

Table 1. Number of schools by school type in the final unified dataset.

Bottom-up analysis approach

Bottom-up analysis referred to a school building selected as a case study for performance modelling. The underlying causes of differences between the model based on design expectations and the calibrated model outputs were investigated and the potential for operational performance of the building was explored. The case building was modelled in DesignBuilder software using the Simple HVAC modelling approach. This approach is based on the National Calculation Method (NCM) activity dataset and predefined HVAC templates, which is one of the modelling approaches proposed in CIBSE TM54 for individual buildings (BRE, 2025; CIBSE, 2022). The modelling results were compared with measured data and the models were validated against the monthly calibration criteria defined in ASHRAE Guideline 14 (ASHRAE, 2014). These criteria use two statistical metrics, the normalised mean bias error (NMBE) and the coefficient of variation of the root mean square error (CVRMSE), to validate that the calibrated model meets acceptable standards (NMBE < ±5% and CVRMSE < 15%). Then feasible energy saving and emission reduction measures were identified by analysing changes in input information and simulation results between the baseline and calibrated models. The bottom-up analysis is used to show an example of the ‘potential’ energy performance of the latest cluster of school buildings covered in the study against the ‘actual’ performance represented in the statistical sample.

Top-down evidence: Differences based on DEC data

Overall school performance

Figure 1 shows the current annual energy use intensity (EUI) of primary and secondary schools in England using cumulative distribution curves. Commonly, the 25th percentile represents buildings with lower energy use and is referred to as the ‘good practice’ benchmark. The 50th percentile represents the median performance and is referred to as the ‘typical practice’ benchmark. (CIBSE, 2021; Hong et al., 2022, 2013). Thus, typical practice for current electricity EUI in primary and secondary schools is 39 kWh/m² and 43 kWh/m² respectively, and 104 kWh/m² and 90 kWh/m² for fossil thermal energy consumption. Mann-Whitney U-tests on the electricity consumption and fossil thermal energy consumption for these two types of schools, respectively, both resulted in p < 0.001, which indicated that there was a statistically significant difference in energy consumption by school type. Secondary schools consumed more electricity and less fossil heat energy than primary schools.

Figure 1. Cumulative distribution curves of annual EUI.

Table 2 summarises the statistics for energy use in primary and secondary schools. Total energy consumption may not necessarily match the sum of electricity and fossil thermal energy consumption because schools with high total energy consumption did not necessarily have the worst performance in electricity and fossil fuel use at the same time.

Table 2.EUI statistics.

Comparison of EUI in three school building stock clusters

Figure 2 shows the annual EUI of the BSF school building stock, the PSBP school building stock, and the remaining school building stock in boxplots for comparison. Within the same building stock clusters, fossil thermal energy use was high in primary schools, while secondary schools consumed more electricity, which is consistent with the results of the overall school energy performance analyses in the previous section. By fuel type, for electricity, BSF secondary schools were the cluster with the highest electricity consumption, PSBP schools consumed more electricity than the remaining schools. For fossil thermal energy, the primary school buildings in the remaining schools consumed the most fossil thermal energy. The secondary school buildings in the PSBP school cluster consumed the least fossil thermal energy. By school type, for primary schools, the PSBP primary buildings had the greatest consumption of electricity, but both fossil thermal energy and total energy consumption were significantly lower than in the remaining schools. For secondary schools, BSF secondary school buildings have the greatest fossil thermal and electrical EUI, and their fossil thermal energy consumption is even higher than that of PSBP primary school buildings. PSBP secondary schools have a slightly higher electrical EUI than the remaining schools, but have the lowest levels of fossil thermal and total energy consumption.

Figure 2. Comparison of EUI in three school building stock clusters.

Bottom-up modelling insights: Revealing energy saving potential

The above study identified PSBP schools as having relatively better operational energy performance by comparing the DEC records of different school clusters. In the following section, a PSBP primary school building, as shown in Figure 3, was used as a case study for a bottom-up analysis of the in-use energy performance at an individual building level. The school, located in the North of England and opened in 2019, is a L-shaped two-storey building with about 2400 m² of floor area. Figure 4 illustrates the building model developed in DesignBuilder software for this case study. Table 3 shows the input parameters and data sources in the baseline model. The baseline model represents the potential operational performance of the building under the design conditions. The information for the modelling was obtained from as-built architectural drawings, building services and engineering schematics collected through site visits. In the meantime, the building’s energy bills for the past year were also collected to get the actual operational energy performance (April 2023 to March 2024). The measured annual natural gas consumption was 81 kWh/m² and the annual electricity consumption was 41 kWh/m², both of which lie between best practice and typical practice for the PSBP primary schools cluster.

Figure 3. Case study school building.

Figure 4. Building model developed for the case study with DesignBuilder.

Table 3. Input parameters for the baseline model.

The baseline model predicted yearly gas and electricity consumption of 21.5 kWh/m² and 30.5 kWh/m² respectively, both of which were better than the good practice PSBP primary school building and put together would meet the RIBA net-zero operational target for schools if achieved in practice (RIBA, 2021). The modelled total energy consumption is in the top 10% of DEC records for this cluster. This, however, presents a significant gap with the in-use energy performance of the building. In order to calibrate the model and identify potential causes of the gap, reasonable scenario assumptions, based on the site visits and a review of the literature of the performance gap in schools (Bordass et al., 2001; Burman et al., 2018; Carbon Trust, 2011; Jain et al., 2020; Jradi, 2020; Pegg et al., 2007),were proposed for testing. Figure 5 shows the scenario test results from the baseline model to the calibrated model and the comparison with the measured data.

Figure 5. Scenario test results from the baseline model to the calibrated model.

Scenario 1 was based on updating the heat gain generated by the equipment in the server room from the NCM value of 50 W/m² to the maximum recommended value of 250 W/m² as per the latest DfE guidelines for PSBP schools (the systematic bias in measured vs. modelled electricity use indicates an underestimation of electrical equipment load in the baseline model)(DfE, 2022). This reflects the impact of the increase in ICT equipment on the electricity consumption. Scenarios 2 and 3 were developed by changing the building background ventilation rate to 0.5 air change per hour (ACH) and extending the operating hours of the heating system during half-term breaks in the heating season. These two scenarios were proposed based on information gathered during the site visit to incorporate random window opening behaviour by occupants and the fact that the heating system would still operate normally during non-term hours into the modelling considerations. For Scenario 4 the heating set point temperature was adjusted upwards from the NCM value to 22°C. The final simulated natural gas consumption was 83.2 kWh/m²/year and electricity consumption was 39.2 kWh/m²/year. This scenario is cumulative of the above scenario assumptions and the calculated NMBN and CVRMSE meet the monthly calibration criteria. The modelling for Scenario 4 is therefore a calibrated model. Figure 6 shows the monthly natural gas and electricity usage simulated by the calibrated model compared to the measured data.

Figure 6. Comparison of simulated monthly natural gas and electricity usage against measured data for the calibrated model.

Discussion

This study found that electricity consumption was higher in secondary schools than in primary schools, while fossil thermal energy consumption was higher in primary schools than in secondary schools. This is consistent with the findings of two previous studies that analysed DEC data prior to June 2012 (Godoy-Shimizu et al., 2011; Hong et al., 2014). However, these two studies showed no statistically significant difference in fossil thermal energy use between primary and secondary schools, which is different from the findings of the current data analysis.Another research on DEC records from 2012-2014 obtained the same result as this study, that is, there was a statistically significant difference in both fossil thermal energy and electricity use in primary and secondary schools (Hong et al., 2022).The potential reason for the significant difference in electricity consumption is that secondary schools have more ICTs and equipment in school facilities.Data on the condition of school buildings collected in both the Property Data Survey Programme (PDSP) and Building Energy Efficiency Surveys (BEES) indicated that secondary schools have more energy-intensive activities and energy-consuming equipment than primary schools (BEIS, 2016; EFA, 2015).The increased internal gain caused by this difference in space and equipment utilisation reduces the need for space heating, which could be a likely reason for the gradual difference in fossil thermal energy use between primary and secondary schools.

For energy use patterns in different clusters of school building stock,the operational energy performance of PSBP schools was relatively better than that of the remaining school building stock, whereas the operational energy performance of BSF schools was worse than that of the remaining school building stock. It is worth noting that each school under the BSF program was designed individually, which was slow and costly to construct. In contrast, PSBP schools adhered to specifically proposed baseline design requirements., reducing project costs by 1/3 compared to BSF schools while improving construction efficiency (DfE, 2015). Besides, under the PSBP program, the space design of primary and secondary schools was reduced by 15% and 5%, respectively, compared to BSF schools, avoiding unnecessary space waste while maintaining equivalent teaching spaces (DfE and EFA, 2012). Although influenced by the updated design requirements in the Building Regulations, both the insulation of the building fabric and the equipment efficiency of the building services systems in the PSBP schools have been improved in comparison to the BSF schools. However, considering the financial and time investment in the BSF program, these modern schools built after 2004 are expected to be more energy efficient than the remaining building stock. These buildings may have inherent operational problems or mismanagement of systems due to complex building design, resulting in the gap between measured performance and design intent.

Bottom-up analyses of the individual building revealed that operational energy performance has a high potential for improvement through enhanced routine maintenance and management of the building services system (e.g. optimum space time utilisation of building services and improving HVAC zoning and schedule arrangements). Thus, a combination of top-down and bottom-up analyses should be used to locate the energy consumption level of the building and to propose feasible energy saving measures. In addition, in order to use building performance modelling for practical purposes, the assumptions for scenario testing need to be based on the actual operation of the building, rather than relying on individually more sensitive input parameters to achieve model calibration.

Conclusion

This paper used the publicly available DEC dataset to report the energy use patterns of primary and secondary schools in England. It also analysed the latest energy consumptionfor different clusters of school buildings. The study found that BSF schools with high financial investment had poor energy performance, while PSBP schools with compact designs had more efficient building operations. However, an investigation of a PSBP primary school at the individual building level still identified significant gaps between the actual operation of the building and the design expectations. And the scenario testing revealed even without refurbishing the building, there were a variety of practical measures that could be taken to further reduce the operational energy consumption of the building. The limitation of this study is that only one building in one cluster was used as a case study, and in the future more cases will be introduced for further comparative studies.

References

Armitage, P., Godoy-Shimizu, D., Steemers, K., Chenvidyakarn, T., 2015. Using Display Energy Certificates to quantify public sector office energy consumption. Build. Res. Inf. 43, 691–709. https://doi.org/10.1080/09613218.2014.975416.

ASHRAE, 2014. ASHRAE Guideline 14-2014 Measurement Of Energy, Demand, And Water Savings.

BEIS, 2016. Building Energy Efficiency Survey (BEES). Department for Business, Energy & Industrial Strategy, London.

Bordass, B., Cohen, R., Standeven, M., Leaman, A., 2001. Assessing building performance in use 3: energy performance of the Probe buildings. Build. Res. Inf. 29, 114–128. https://doi.org/10.1080/09613210010008036.

BRE, 2025. NCM: iSBEM for England [WWW Document]. URL https://www.uk-ncm.org.uk/download.jsp?id=35 (accessed 3.21.24).

Burman, E., Kimpian, J., Mumovic, D., 2018. Building Schools for the Future: Lessons Learned From Performance Evaluations of Five Secondary Schools and Academies in England. Front. Built Environ. 4, 22. https://doi.org/10.3389/fbuil.2018.00022.

Carbon Trust, 2011. Closing the gap: Lessons learned on realising the potential of low carbon building design. Carbon Trust, London, UK.

CIBSE, 2022. TM54: Evaluating operational energy use at the design stage. The Chartered Institution of Building Services Engineers, London.

CIBSE, 2021. Energy Benchmarking Tool [WWW Document]. URL https://www.cibse.org/knowledge-research/knowledge-resources/knowledge-toolbox/benchmarking-registration (accessed 10.10.24).

CIBSE, 2008. TM46: Energy benchmarks, Technical Memoranda. The Chartered Institution of Building Services Engineers, London.

DCLG, 2012. The Energy Performance of Buildings (England and Wales) Regulations 2012.

DCLG, 2007. The Energy Performance of Buildings (Certificates and Inspections) (England and Wales) Regulations 2007.

DfE, 2024. Get Information about Schools - GOV.UK [WWW Document]. URL https://get-information-schools.service.gov.uk/ (accessed 8.29.24).

DfE, 2022. School Output Specification Technical Annex 2H: Energy.

DfE, 2021. Major rebuilds begin at schools across England [WWW Document]. GOV.UK. URL https://www.gov.uk/government/news/major-rebuilds-begin-at-schools-across-england (accessed 1.23.22).

DfE, 2015. £6 billion investment to rebuild and improve schools across England [WWW Document]. GOV.UK. URL https://www.gov.uk/government/news/6-billion-investment-to-rebuild-and-improve-schools-across-england (accessed 1.14.25).

DfE, 2012. Equalities impact assessment: building schools for the future.

DfE, EFA, 2012. Innovative new school designs deliver efficiency for every pound spent [WWW Document]. GOV.UK. URL https://www.gov.uk/government/news/innovative-new-school-designs-deliver-efficiency-for-every-pound-spent (accessed 1.14.25).

DLUHC, 2024. Energy Performance of Buildings Data England and Wales [WWW Document]. URL https://epc.opendatacommunities.org/domestic/search (accessed 9.23.24).

Education and Skills Committee, 2007. Sustainable Schools: Are we building schools for the future? Seventh Report of Session 2006–07. House of Commons: Education and Skills Select Committee, London.

EFA, 2015. Property data survey programme. Education Funding Agency, London.

ESFA, 2016. Priority School Building Programme: overview.

Godoy-Shimizu, D., Armitage, P., Steemers, K., Chenvidyakarn, T., 2011. Using Display Energy Certificates to quantify schools’ energy consumption. Build. Res. Inf. 39, 535–552. https://doi.org/10.1080/09613218.2011.628457.

Hong, S., Mylona, A., Davies, H., Ruyssevelt, P., Mumovic, D., 2018. Assessing the trends of energy use of public non-domestic buildings in England and Wales. Build. Serv. Eng. Res. Technol. https://doi.org/10.1177/0143624418806135.

Hong, S.M., Godoy-Shimizu, D., Schwartz, Y., Korolija, I., Mavrogianni, A., Mumovic, D., 2022. Characterising the English school stock using a unified national on-site survey and energy database. Build. Serv. Eng. Res. Technol. 43, 89–112. https://doi.org/10.1177/01436244211030667.

Hong, S.-M., Paterson, G., Burman, E., Steadman, P., Mumovic, D., 2013. A comparative study of benchmarking approaches for non-domestic buildings: Part 1 – Top-down approach. Int. J. Sustain. Built Environ. 2, 119–130. https://doi.org/10.1016/j.ijsbe.2014.04.001.

Hong, S.-M., Paterson, G., Mumovic, D., Steadman, P., 2014. Improved benchmarking comparability for energy consumption in schools. Build. Res. Inf. 42, 47–61. https://doi.org/10.1080/09613218.2013.814746.

Jain, N., Burman, E., Robertson, C., Stamp, S., Shrubsole, C., Aletta, F., Barrett, E., Oberman, T., Kang, J., Raynham, P., Mumovic, D., Davies, M., 2020. Building performance evaluation: Balancing energy and indoor environmental quality in a UK school building. Build. Serv. Eng. Res. Technol. 41, 343–360. https://doi.org/10.1177/0143624419897397.

James, S., Brown, E., 2013. Review of education capital: progress update (No. DFE-00286-2013). Department for Education.

Jradi, M., 2020. Dynamic Energy Modelling as an Alternative Approach for Reducing Performance Gaps in Retrofitted Schools in Denmark. Appl. Sci. 10, 7862. https://doi.org/10.3390/app10217862.

Mann, H.B., Whitney, D.R., 1947. On a Test of Whether one of Two Random Variables is Stochastically Larger than the Other. Ann. Math. Stat. 18, 50–60. https://doi.org/10.1214/aoms/1177730491.

Pegg, I.M., Cripps, A., Kolokotroni, M., 2007. Post-occupancy performance of five low-energy schools in the UK. ASHRAE Trans. 113, 3–14.

RIBA, 2021. RIBA 2030 Climate Challenge.