Comparing different approaches of agent-based occupancy modelling for predicting realistic electricity consumption in office buildings

Mashuk, Md Shadab; Pinchin, James; Siebers, Peer-Olaf; Moore, Terry

Abstract

Having a good grasp on modelling the dynamics of occupants for estimating electricity consumption in office buildings is a vital asset for realistic predictions. Nowadays, agent-based models are widely used for this purpose. Previous approaches to modelling dynamics of occupants in multi-floor office buildings simplified the models by teleporting agents between zones during transitions without considering the routes used to reach their final destination such as going through corridors, stairways and hallways, thus, underestimating the potential energy usage during those transition period. This paper proposes a more realistic approach by incorporating detailed routes of agent movement when transiting from one zone to another. To demonstrate the case, detailed routes and route choice preferences are used as inputs within the model for the agents to make independent decisions when transiting from one place to another within the simulated office building. The route choice preferences are computed from data gained from an earlier extensive real world occupancy detection trial conducted within the model office building using state of the art indoor positioning system. The simulation experiments compare the previous approach against the proposed approach and based on the evaluation it is found that there is approximately 19% underestimation of electricity consumption per day when detailed routes are not considered. The research demonstrates, the proposed approach is applicable to any office buildings and will produce predictions which will be much more realistic and closer to the real world electricity consumption level.

Citation

Mashuk, M. S., Pinchin, J., Siebers, P., & Moore, T. (2024). Comparing different approaches of agent-based occupancy modelling for predicting realistic electricity consumption in office buildings. Journal of Building Engineering, 84, Article 108420. https://doi.org/10.1016/j.jobe.2023.108420

• Journal Article Type: Article
• Acceptance Date: Dec 29, 2023
• Online Publication Date: Dec 31, 2023
• Publication Date: May 1, 2024
• Deposit Date: Jan 8, 2024
• Publicly Available Date: Jan 25, 2024
• Journal: Journal of Building Engineering
• Electronic ISSN: 2352-7102
• Publisher: Elsevier
• Peer Reviewed: Peer Reviewed
• Volume: 84
• Article Number: 108420
• DOI: https://doi.org/10.1016/j.jobe.2023.108420
• Keywords: Building occupancy, Building performance, Indoor positioning, Agent based modelling, Electricity consumption
• Public URL: https://nottingham-repository.worktribe.com/output/29273060