An Agent-Based Occupancy Simulator for Building Performance Simulation

Traditionally, in building energy modeling (BEM) programs, occupancy inputs are deterministic and less indicative of real world scenarios, contributing to discrepancies between simulated and actual energy use in buildings. This paper presents an agent-based occupancy simulator, which models each occupant as an agent with specified movement events and statistics of space uses. To reduce the amount of data inputs, the simulator allows users to group occupants with similar behaviors as an occupant type, and spaces with similar function as a space type. It is a web-based application with friendly graphical user interface, cloud computing, and data storage. A case study is presented to demonstrate the usage of the occupancy simulator and its integration with EnergyPlus and obFMU. It first shows the required data inputs and the results from the occupancy simulator. Then, the generated occupant schedules are used in the EnergyPlus and obFMU simulation to evaluate the impacts of occupant behavior on building energy performance. The simulation results indicate that the occupancy simulator can capture the diversity of space’s occupancy behavior rather than the static weekly profiles, and can generate realistic occupancy schedules to support building performance simulation. INTRODUCTION Traditionally, in building energy modeling (BEM) programs, occupancy inputs are deterministic and less indicative of real world scenarios, contributing to discrepancies between simulated and actual energy use in buildings. The International Energy Agency (IEA) Energy in the Buildings and Communities Program (EBC) Annex 53 (Total Energy Use in Buildings: Analysis & Evaluation Methods) pointed out that occupants’ activities and behavior is one of the six key factors directly influencing building energy use. Occupant behavior is now widely recognized as a major contributing factor to uncertainty of building performance (Yan et al., 2015). The occupant behaviors can be organized into two categories: occupancy and occupants’ interactions with building systems (Wang et al., 2011). The occupancy determines the location of each occupant during each time period. When occupants are located in a space, they may be able to control the building systems (such as lights, HVAC, and windows), and therefore impact the energy consumption. The occupancy simulation is the foundation of occupant behavior research. Wang et al. (2011) introduced a novel approach for building occupancy simulation based on Homogeneous Markov chain model, which simulates the stochastic movement of occupants. The model can generate location for each occupant and the occupancy of each space of a building. The model was implemented using a mathematical analysis software package as a proof of concept. Later on, Feng et al. (2015) updated and implemented the model in C++, an object-oriented programming language. The software module can be used as a stand-alone application to pre-calculate occupancy schedules, or be integrated into building energy modeling programs as a dynamic link library. Hong et al. (2015a) further integrated the occupancy simulation model into the occupant behavior functional mockup unit (obFMU) for co-simulation with BEM programs. The above mentioned occupancy simulation tools are all desktop applications, which are hard to maintain and only compatible