The impact of occupant behaviour in dwellings on building energy policy: integrating diverse occupancy distributions into the UK standard assessment procedure

The UK Government is committed to achieving net-zero greenhouse gas emissions by 2050. To achieve this, the UK dwelling stock, responsible for 20% of UK emissions in 2021, will need to be decarbonised. Energy demand policy goals such as the decarbonisation of heat in buildings will require large-scale transformation of energy use in dwellings.

Energy modelling tools provide a method of identifying the effectiveness, cost, and impact of policy initiatives to reduce energy demand and emissions within the built environment. However, uncertainty in many of the assumptions used in modelling tools can significantly impact the reliability of model estimations. The Standard Assessment Procedure plays an integral role in the evaluation of energy demand policy initiatives, previously used to underpin schemes such as the Green Deal and Renewable Heat Incentive. 

The Standard Assessment Procedure uses standardised and deterministic assumptions for inputs including occupant behaviour. However, literature has identified that the behaviour of occupants is influential in the energy demand of dwellings. Literature has also determined that the use of standardised assumptions can impact the accuracy of energy demand estimations, which affects the reliability of policy initiatives.

This work sets out to investigate the influence of occupant behaviour on the predictions of energy modelling tools, focussing on the Standard Assessment Procedure (SAP). This is achieved by introducing probability distributions into SAP in place of the standard assumptions currently used for occupant behaviour. This enables the level of uncertainty in the total annual energy demand estimations of SAP to be quantified.

Firstly, the work provides a sensitivity analysis of the SAP model to identify the parameters that most influence energy demand. Then, an uncertainty analysis is performed to quantify the level of uncertainty in energy demand estimations due to the diversity in occupant behaviour. Finally, this approach is applied to the evaluation of an example energy demand policy initiatives to illustrate how the outcome of policy could be affected by the way the behaviour of occupants is modelled.  

Probability distributions for thermostat setpoint, heating schedule, household size, daily hot water usage and lighting demand were introduced to SAP. This work has identified that energy demand estimations varied by up to ±50% from a standard SAP estimation and were on average 17-20% below the standard SAP estimation. For 75% of occupancy scenarios, energy demand was below the standard SAP estimation implying that SAP overestimates total annual energy demand for the majority of occupants.

Applying the diverse occupancy distributions to an example energy demand policy identified that for 52% of households in a section of the English dwelling stock (terraced, semi-detached, detached dwellings), transitioning from a gas boiler to an air-source heat pump would reduce total annual fuel bills (not including electrical appliances), reducing to 12% of households if a change to continuous heating is assumed. For all SAP runs, overall household cost is higher over 15-years for an air-source heat pump than a gas boiler due to the significantly higher upfront cost to install an air-source heat pump. Introducing cost-effective retrofit measures (e.g., loft insulation, cavity wall insulation, draught proofing, floor insulation) before installing a heat pump offset the higher household cost for up to 35% of households.

This work has demonstrated that uncertainty levels due to occupant behaviour within energy demand predictions can be useful in improving the reliability of energy demand estimations for dwellings and providing insight into the proportion of households where an intervention is beneficial, which could be useful in identifying the risks that energy demand policy may not perform well.