Validation of a meso-scale Urban Building Energy Model of domestic peak heating load and annual heat demand in England and Scotland

This article presents a methodology to model domestic peak heating load and annual heat demand at the urban scale using geospatial data and building stock data from Energy Performance Certificates (EPC). The model was run on 720269 domestic properties across three different local authorities in Scotland (Aberdeen City, Dundee City Council and Fife) and two combined authorities in England (Somerset Council and West Midlands Combined Authority).

Then, the results were validated on a mean absolute percentage error (MAPE) basis against smart meter gas demand readings from 4397 properties included in the Smart Energy Research Lab (SERL) Statistical data set. The model was able to predict peak heating load with a 4% MAPE in F-rated properties in Fife and annual heat demand with a 6% MAPE in A-rated properties in Somerset Council. However, the reported MAPE goes as high as 70% for peak heating load and 57% for annual heat demand. Further validation on an aggregated basis is required against a comprehensive data set that includes separate gas demand readings for space and hot water heating.