Affordable Access

Harvesting the low-hanging fruit of high energy savings -- Virtual Occupancy using Wi-Fi Data

Authors
  • Clark, Callie
  • Prakash, Anand
  • Pritoni, Marco
  • Kloss, Margarita
  • Gupta, Pranav
  • Nordman, Bruce
  • Piette, Mary Ann
  • Kamel, Michael
  • Semaan, Tony S
  • Eisele, Ann
  • Hage, Dotty
  • Flannery, Pat
Publication Date
Oct 15, 2020
Source
eScholarship - University of California
Keywords
License
Unknown
External links

Abstract

Approximately 20% of primary energy consumed in the U.S. is attributed to HVAC use. Ideally, HVAC operation would be driven by actual building occupancy, but lack of reliable occupancy information often results in the use of conservative static schedules. This disparity is even more pronounced in a college campus, where the function of each space differs by building (classrooms, offices, libraries) and the class schedules change frequently -- every semester, day of week, and hour. While several research papers propose the use of counts of the Wi-Fi connections (e.g., phones, computers) as a proxy for occupancy, few real-world implementations exist. This paper describes the development and deployment of an open-source Wi-Fi-to-Occupancy software library in 65 buildings of a college campus, and the planned integration with the building energy management and control system at the building scale. Over a year of Wi-Fi data was gathered into distinct academic periods, including fall and spring semester, academic breaks, and summer sessions. Patterns such as students moving between classrooms, closing laptops before exams, etc., can be visualized from the data. Approximating occupancy from Wi-Fi data presents challenges which we address in this project -- for example, identifying static devices, or estimating the ratio of devices per person. Utilizing real-time occupancy data to inform optimal HVAC schedules and ventilation rates creates the potential to identify and reduce energy waste. Other potential applications include forecasting occupancy, and using Wi-Fi data to predict peak demands. Finally, the paper discusses how to easily scale these tools to other buildings.

Report this publication

Statistics

Seen <100 times