XClose

UCL Institute for Environmental Design and Engineering

Home
Menu

Duncan Grassie

An investigation of feedback and feed-forward energy efficiency mechanisms from a UK school crowdsourced building stock model

Novel methods of modelling energy demand in buildings are required to assist with tracking progress towards the UK government’s commitment to Net Zero by 2050. The development of automated methods, combined with national datasets in building construction and geometry, have facilitated the construction of 16,000 UK school EnergyPlus building simulation models which can be used to run scenarios for refurbishment. These models successfully integrate the top-down comparison of schools’ headline energy ratings relative to peers, with bottom-up component by component analysis based on building physics. However, a key issue identified in my preceding MRes project,  was that building service equipment, schedules and setpoints datasets (heating, lighting, IT, catering, etc.) are also necessary to differentiate between poorly fabricated and poorly operated school buildings.

Previous efforts to determine building energy performance has involved engaging the engineers responsible for designing the buildings, however, head teachers, facilities managers, bursars and teachers themselves are more aware of the actual day to day operation of the building. Complex monitoring and post occupancy evaluations projects can be carried out up to a point with budget and personnel constraints limiting the scale. However, large scale modelling projects will require engagement of these key groups by helping to overcome knowledge or motivational gaps to crowdsource this building services data.

This multi-disciplinary project investigates a number of facets of a successful data crowdsourcing method:

  • Stakeholder workshop/interview sessions to determine key drivers, decisions and data requirements for groups operating and working with energy performance of school buildings
  • Data analysis of the CarbonBuzz crowdsourced dataset to determine availability and reliability of key variables used in modelling
  • Sensitivity analysis to determine the criticality of a number of key variables in stock modelling.

The second half of the projects tests various design aspects of a crowdsourcing method:

  • Identification of practical difficulties with adopting the National Calculation Methodology as the standard template for operation of school buildings
  • Implementation of a pilot crowdsourcing project in London schools testing response rate and engagement
  • Statistical analysis and aggregation of data obtained by crowdsourcing to a stock level

Much of this work is expected to be published in the second half of 2020 – for more details please visit my repository at UCL Discovery.