Paper Conference

Proceedings of eSim 2018: 10th Conference of IBPSA-Canada

     

On the evaluation of occupancy models at various spatial scales

Sara Gilani, William O’Brien

Abstract: Occupancy data is an essential input in occupant studies to predict occupants-building interactions and to automate building controls. Therefore, it is important to find an occupancy modelling approach that can better predict occupancy. While previous research evaluated the performance of occupancy models for individual offices, such evaluation at varying zone sizes is required as well. This research investigates the deviations between measured and predicted occupancy data at various spatial resolutions. Occupancy data were collected in 24 offices in an academic building in Ottawa, Canada. Measured data were compared with predicted data which were estimated by three occupancy models, including an existing occupancy model, a custom schedule-based model, and a classification decision tree-based model. At different spatial scales, using random sampling from the individual monitored offices for the considered number of zones, the deviations between measured and predicted results were calculated. This analysis indicated that the performance of the considered occupancy models varied from one occupancy-related measure to another at different spatial scales. However, the errors of the predicted measures approached consistent values at a given number of offices. For this case study, the existing occupancy model and custom schedule-based model showed better performance compared to the classification decision tree-based model.
Keywords: Occupants’ presence; Occupancy modelling approach; Spatial resolutions; Measurement and prediction; Evaluation.
Pages: 404 - 412
Paper:
esim2018_2-2-A-3