A novel model to detect abnormal thermal energy consumption in existing buildings with limited metering infrastructure

In EU, excessive energy waste occurs in existing buildings due to multiple reasons, such as inadequate building design, frequent operation of the technical building systems far from the design and reference conditions, lack of proper maintenance, lack of building automation and control systems, unexpected or exceptional weather conditions and inappropriate end-user behaviours. In this context, the detection of “abnormal” energy consumption is crucial both to increase end-users’ awareness and to help energy managers and technicians in the energy diagnosis. In this paper, the authors present a novel model to detect abnormal thermal energy consumption in existing buildings equipped with limited metering infrastructure. Four heating seasons of daily thermal energy consumption and indoor environmental data of a case study building located in Central Italy were analysed to develop, train and test the model for data processing, benchmarking, abnormal energy consumption identification and diagnosis. The results of the test phase show the ability of the model to detect behavioural faults, such as the low consumption and indoor air temperature in two dwellings and the high ones in one other. Also, the poor thermal insulation properties of the building were clearly highlighted by the proposed model.