Paper Conference

Proceedings of uSim Conference 2020: 2nd uSim Conference of IBPSA-Scotland


Control event matrices for stochastic heating use in urban-scale energy simulations

Peter McCallum, David Jenkins, Paraskevi Vatougiou

Abstract: The prevalence of Urban-scale Building Energy Modelling (UBEM) in recent literature emphasises the significant shift of focus in how we consider energy demands from the built environment. This change has of course been driven by the major transitions taking place in our energy systems, resulting from far-reaching decarbonisation initiatives. As is always the case for numerical modelling, a model is only as good as the data used in its formation. This work, contributes to one facet of the growing basis of knowledge around formal Uncertainty Analysis in building modelling – the aleatory uncertainty associated with the stochastic behaviour of people. A hybrid rule-based, clustering method is presented, which identifies events in households that correspond to a changes between setback and comfort temperatures. To catalogue the resulting data, ‘control event matrices’ are compiled to record the probabilistic relationships between the timings of set point changes. This particular scheme has been designed to provide a lightweight and extensible model; it facilitates both the automated processing of large scale smart meter datasets, and the stochastic regeneration of control schedules for UBEM. The demonstration data (AECOM, 2018) featured 1,300 dwellings, and resulted in data structures five orders of magnitude smaller than the original dataset. The basis for generating such data is to provide a methodology for interpreting, anonymising and condensing vast records of smart meter data, to potentially offer near realtime observation of widespread behavioural change around heating practices and technology adoption, and augment traditional Time Use Survey methods.
Pages: 204 - 211