W8UP3 Impact of data availability and model complexity on prediction of energy consumption in Camden schools
5th June 2017 Duncan Grassie

Impact of data availability and model complexity on prediction of energy consumption in Camden schools

Duncan Grassie, UCL Energy Institute


Energy demand in non-domestic buildings is responsible for around a fifth of the UK total’s carbon emissions (1) and so these buildings are key to the UK meeting its 2050 commitment (2) to reduce carbon emissions. School buildings are a useful testing ground for non-domestic energy efficiency measures; as buildings under national ownership and control (3), geometry and fabric datasets exist at a national level, facilitating the auto construction of building simulation models of the entire school stock in EnergyPlus through UCL’s SimStock tool (4).

Such stock models combine physics based insights of which building services may be responsible for energy demand but across a population level so could be used by policy makers to predict or track future performance and target measures at individual buildings or sectors within the stock. However a performance gap (5) between model predicted performance with measured annual performance as indicated by Display Energy Certificates (DECs) can reduce the effectiveness of measures in practice.

A case study of three complex Camden school campuses using the SimStock modelling approach was carried out to demonstrate which factors may be responsible for the gap. Methods were developed for automating fabric related glazing ratios and defining building types based on age. Sensitivities were carried out on weather, geometry, fabric and occupant schedules and setpoints, accounting for both the potential range of corrections as well as the effect on overall energy consumption.

It was demonstrated that heating requirements were significantly over predicted when using generic classroom environmental guidelines. Corrections due to non-angular buildings and modern extensions to older buildings were found to be largely insignificant relative to changes in heating setpoints, demonstrating a need for a UK-wide school dataset in operating setpoints and schedules. A method of Crowdsourcing this data directly from school users was proposed as the next stage following this Masters project, which will be tested in the PhD itself.


  1. Department of Energy and Climate Change (DECC), 2008, “Climate Change Act 2008, Part 1”, UK Government.
  2. The Carbon Trust, 2009, “Building the future, today – Transforming the economic and carbon performance of the buildings we work in”, The Carbon Trust.
  3. Pereira, D. Raimondo, S. P. Corgnati, and M. Gameiro da Silva, 2014, “Energy consumption in schools: A review paper,” Renew. Sustain. Energy Rev., vol. 40, pp. 911–922, Dec. 2014.
  4. Coffey, B., Stone, A., Ruyssevelt, P., Haves, P., 2015, “An Epidemiological Approach to Simulation-Based Analysis of Large Building Stocks”, Proceedings of IBPSA International. Hyderabad.
  5. Demanuele, C., Tweddell, T., Davies, M., 2010, “Bridging the gap between predicted and actual energy performance in schools”, Proceedings of World Renewable Energy Congress XI, Abu Dhabi. 

Project Team

Duncan Grassie

Paul Ruyssevelt
Ivan Korolija


MRes Conference – Grassie

Title Link


Presented at the Bartlett Conference at UCL