Conference Paper
Proceedings of eSim 2022: 12th Conference of IBPSA-Canada
Cluster-Based Surrogate Modelling Approach to Remote Building Characterization
Cluster-Based Surrogate Modelling Approach to Remote Building Characterization
Shane Ferreira 1, Burak Gunay 1, Brent Huchuk 2, Scott Shillinglaw 2, Farzeen Rizvi 21 Carleton University, Canada2 National Research Council CanadaAbstract: To better inform retrofit decision-making and benchmarking, a scalable characterization method needs to be developed. The conventional field-scale and inverse model-based characterization approaches are limited in their scalability due to their engineering cost and need for high fidelity building automation system (BAS) data. To this end, a cluster analysis-based method is developed to estimate the thermophysical features of the envelope, HVAC operation, and casual heat gains of a building based on heating and cooling load profiles. Variants of a mid-rise office building energy model were generated through random sampling of these features. Then, for each variant, three-parameter univariate heating and cooling change-point models were fitted separately for operating and afterhours. Different clustering techniques were applied to the change point model parameters of all variants. Building feature sets that are likely associated with each cluster are identified. The method could effectively identify energy-related building features from heating and cooling load profiles. Keywords: Surrogate Modelling, Clustering, Building CharacterizationPaper:![]()
