Occupant behaviour modeling for building energy conservation : an integrated approach using agent based, system dynamics and building information modeling

Abstract

Energy consumption in buildings is affected by various aspects, including its physical characteristics (i.e., Interior layout, orientations, etc.), the appliances inside (i.e., HVAC, light, other devices.), and the ambient environment. However, the occupant's behaviour that determines and regulates the building energy consumption must not be forgotten. In most previous researches and simulation tools (i.e., EnergyPlus, e-Quest, etc.), occupant behaviour is modelled as static or fixed occupancy profiles. These profiles are acknowledged as the main source of discrepancy between the predicted and actual building energy performance. Therefore, researchers attempt to model occupants' presence, movement, and adaptive actions more realistically. Based on the identified knowledge gap, this research focuses on comprehensive energy behaviour representation in reviewing the occupant perception, attitude, and behaviour mechanisms considering the Theory of Reasoned Action (ToRA) model. It introduces a new hybrid modelling approach using an Agent-Based Modelling (ABM), System Dynamics (SD), and Building Information Modeling (BIM) that helps to predict the occupant stochastic energy consumption behaviours and indoor ambient parameters in the existing buildings. Moreover, it also calculates the existing building indoor layout impact on energy conservation behaviour through the implementation of Enablement intervention (i.e., Interior Layout deployment). Thus, the primary aim of this research is to develop an integrated framework between the ABM-SD and BIM, which is capable of analyzing and prompting the building's energy conservation with improved accuracy by considering the dynamic influencing factors through an interdependent analysis. In line with this aim, five objectives were set. i) To identify the theoretical framework of energy consumption behavior and mutual factors (i.e., social and factors related to building and layout) involved in building energy conservation due to dynamic human behavior (Literature Review). ii) To develop an integrated (ABM-SD-BIM) model that appraises and investigates the various energy consumption events with the variation of indoor parameters contributing to occupants' satisfaction (Java-based modelling tools: AnyLogic, Revit Dynamo). iii) To appraise the comprehensive energy-related behavior determinants (i.e., psychological and physiological) and monitor the stochastic behavior pattern for the building occupants (From model). iv) To investigate the influence of interior layout deployment (i.e., placement of stuff) on the building energy conservation under a contextual intervention (i.e., Enablement) for an individual and group of occupants from low-income economies (i.e., Bangladesh). v) To validate the hybrid model using real data (e.g., customized sensors) and paper-based surveys to check the model performance and improve the energy conservation events. The validation study has been conducted to test the behavior hybrid model with the visualization techniques and calculation of confusion metrics for model performance evaluation (i.e., Black-box approach). As key outcomes, the hybrid model actively predicts the stochastic occupant presence and movement, comfort level, energy consumption patterns, temperatures, and CO2 concentration in the indoor space. Moreover, it has been shown that the interior layout adjustment (i.e., intervention) can improve the building energy performance by 14.9%. In terms of energy data validation, the proposed hybrid model has been shown an acceptable range of accuracy with an average CV(RMSE) =10.5%, MBE=1.5%, and R2=0.77. In addition, referring to the confusion matrix, the proposed model has demonstrated exemplary performance as the average predictions reached a relatively good performance, approximately 70%-90%. This study adds another contribution to the existing occupant behaviour research and building energy optimization for enhanced simulation performance. The proposed hybrid model differs from other available studies in two prospects. Firstly, the model adopts an interior layout-based human behaviour study that considers the stochastic occupant attitudes and subjective norms. Secondly, the model is created together with the intervention and consequent validation study to promote energy savings. Thus, the study will help develop a flexible and comprehensive dynamic simulation platform to study both energy-efficient building design and occupant well-being.