The power temperature gradient (PTG) method for deriving an energy signature (a set of parameters that describe the in-use energy performance of a building) is applied to a large smart meter dataset. The resulting parameters are shown for British homes (n = 3899) over a five-year period. Although these parameters are strongly influenced by the buildings’ physical properties, variation in occupant behaviour in buildings that have not been physically modified can also be detected. The median heating power loss coefficient (HPLC) in 2023/24 is 198 W/K. HPLC drops from 2020 to 2022, then rises in two subsequent years. The same trends over time are seen across all household income bands. Smart meter data from 8,880 homes in the Smart Energy Research Lab (SERL) Observatory in the period 2023–24 are used to investigate the strengths and limitations of different variations of the PTG model. The median balance temperature (outside temperature below which space heating is used) for the dataset (15.4°C) is shown to be close to the UK degree-day assumption of 15.5°C. Daily demand data from smart meters enable a higher level of confidence in the energy signature results compared with monthly aggregated data. POLICY AND PRACTICE RELEVANCE Energy signatures provide information for policymakers on how a combination of physical and behavioural factors results in changes in the energy demand of homes. These are distinct from building fabric performance metrics in that they indicate the actual, in-use energy demand. This is important for tracking changes in demand and evaluating progress towards achieving building efficiency targets. The method can be used to assess how interventions affect energy demand in practice, and to distinguish between changes in fabric thermal efficiency, appliance efficiency and heating practices. With the increasing availability of smart meter data, the PTG analysis provides a straightforward method for researchers and energy professionals to characterise in-use building energy demand remotely. Guidance is provided about the strengths and limitations of different variations of the method.
Hanmer et al. (Thu,) studied this question.