Simulating the human-building interaction: Development and validation of an agent-based model of office occupant behaviors

Authored by Jin Wen, Jared Langevin, Patrick L Gurian

Date Published: 2015

DOI: 10.1016/j.buildenv.2014.11.037

Sponsors: United States National Science Foundation (NSF)

Platforms: MATLAB

Model Documentation: ODD Flow charts Ontologies Mathematical description

Model Code URLs: Model code not found

Abstract

This paper develops and validates an agent-based model (ABM) of occupant behavior using data from a one-year field study in a medium-sized, air-conditioned office building. The full ABM is presented in detail using a standard protocol for describing this type of model. Simulated occupant ``agents{''} in the full ABM behave according to Perceptual Control Theory, taking the most immediate, unconstrained adaptive behaviors as needed to maintain their current thermal sensation within a reference range of seasonally acceptable sensations. ABM validation assigns simulated agents the personal characteristics and environmental context of real office occupants in the field study; executes the model; and compares the model's ability to predict observed fan, heater, and window use to the predictive abilities of several other behavior modeling options. The predictive performance of the full ABM compares favorably to that of the other modeling options on both the individual and aggregate outcome levels. The full ABM also appears capable of reproducing more familiar regression relationships between behavior and the local thermal environment. The paper concludes with a discussion of the model's current limitations and possibilities for future development. (C) 2014 Elsevier Ltd. All rights reserved.

Tags

Protocol Predict Thermal sensation Energy-consumption Roc analysis Comfort Windows Fans