Investigating building stock energy and occupancy modelling approaches for district-level heating and cooling energy demands estimation in a university campus
Salam Al-Saegh, Vasiliki Kourgiozou, Ivan Korolija, Rui Tang, Farhang Tahmasebi, Dejan Mumovic
Abstract
• Comparative study of five modelling scenarios for building stock energy demand estimation. • Novel Energy Data-Driven Occupancy Schedule (EDDOS) method introduced. • Increased occupancy modelling complexity improved peak heating load estimation accuracy • Multiple occupancy patterns approach balances accuracy and computational efficiency and the need for real data availability. • Significant variations in cooling demand estimates between different modelling approaches. The urgency of decarbonizing the built environment requires precise modeling of building stock energy performance for effective large-scale planning and retrofitting. Despite advancements in data and modeling techniques, uncertainties persist in balancing model complexity and accuracy, especially in representing occupancy patterns and their impact on energy demand at district and urban scales. This study examines various approaches to building stock energy simulation and occupancy modeling for district-level heating and cooling energy demand, using 19 buildings at a Central London campus as a case study. Five scenarios were evaluated: Scenario A employs THERMOS, a data-driven approach; Scenario B uses a single dynamic thermal simulation model for the entire inventory; Scenario C applies a thermal model with a uniform occupancy schedule across all buildings; Scenario D uses a thermal model with five distinct occupancy profiles; and Scenario E assigns unique occupancy profiles based on energy use data. Results showed that Scenario E’s annual heating demand estimation closely matched metered data (12 % difference), while Scenario A underestimated by 44 %. Complex occupancy models improved peak heating load predictions, with Scenario E showing only a 4 % difference from metered data, though it may not always be feasible due to data and computational constraints. Scenario D emerged as a promising balance between accuracy and efficiency. For cooling demand, significant differences among scenarios (56.43 to 6.1 kWh/m 2 /Y) underscored the importance of accurate occupancy modeling. This research identifies the optimal balance between model complexity and prediction accuracy, introduces the Energy Data-Driven Occupancy Schedule (EDDOS) method, and highlights the potential of data-driven approaches to enhance energy demand assessments.