Clustering-based agent system (CAS) to simulate the energy-related behaviours of office occupants

Abstract

Rapid urbanization and building sector growth emphasize the critical role of energy conservation in addressing global energy consumption and greenhouse emissions. Despite advancements in energy-efficient technologies, an ‘energy performance gap’ exists between predicted and actual energy use, significantly influenced by occupant behaviour. This study explores energy-related behaviour in office buildings by integrating existing behavioural theories including the Theory of Planned Behaviour and the Self-determination Theory, and construct of habit and comfort. Data from an online survey were analyzed using principal component analysis, two-step cluster analysis, and descriptive statistics, identifying three behavioral clusters: ‘Cautious Saver’, ‘Compelling Dissatisfied’, and ‘Coherent Potent’. These clusters represent distinct energy-related behaviours. A Clustering-based Agent System (CAS) was then proposed to simulate the energy-related behaviours of these clusters, offering a dynamic and adaptive modelling framework. The study advocates for a comprehensive approach, integrating behavioural theories to provide insights for developing accurate occupant behaviour models.