The study, published in the April issue of the journal Energy and Buildings, examines data from MIT’s buildings 37 and E52, and finds that while electricity use corresponds to occupancy fairly well in those spaces, the activity of HVAC systems in the buildings does not correlate closely to occupancy.

“It can take a huge amount of energy to heat up buildings during the day, and then suddenly there may be nobody there,” said co-author Carlo Ratti, director of the Senseable City Laboratory and associate professor of the practice in MIT’s Department of Urban Studies and Planning (DUSP). “There is significant potential for improvement and savings, if you could get a more dynamic usage of energy in this area.”

As Ratti and the co-authors note, the federal government estimates that commercial buildings account for about 20 percent of U.S. energy consumption, and 12 percent of U.S. contributions to greenhouse gas emissions; past studies have shown that commercial buildings could reduce their energy use 20 to 30 percent by implementing strategies that better match energy use and need.

The new study uses data about Wi-Fi connections as a proxy for building occupancy, a method the researchers believe could be replicated elsewhere at low cost. While the data does not necessarily reveal an exact population count inside buildings, it does indicate relative occupancy levels over time. “It’s a way of trying to make use of information that already exists,” Ratti said.

The researchers analyzed two different kinds of buildings. Building 37 at MIT houses researchers in astrophysics, aeronautics, and astronautics in a combination of offices, classrooms, and labs. By contrast, Building E52 is the long-time home of MIT’s Department of Economics, with a sizable entrance atrium and a large number of offices inside.

Analyzing data from all four seasons of 2006, the researchers found that Building E52 has lower levels of energy use, and that both buildings have a distinctive cyclical “signature” of electricity usage that rises and falls daily. Both buildings use more steam (for heat) in winter and spring, and more chilled water (for air conditioning) in summer and fall. But while about two-thirds of the variation in electricity levels can be accounted for by changing occupancy levels, the use of the HVAC systems correlated only weakly to occupancy.

In short, these MIT buildings tended to be heated or cooled over extended periods of time according to season, but not in a way that optimized the use of energy. “We were not surprised to see this disconnect, but it was good to be able to quantify it,” said Prudence Robinson, a researcher in the Senseable City Lab and a co-author of the paper.

There are many possible architectural or engineering solutions to this issue, Ratti noted. The larger point, he said, is that “you want to have a better match. You can move the people to the energy, in which case the architecture can help a lot, or you can move the energy to the people, which is more futuristic.” New sensing-based thermostats, for instance, could regulate temperatures on a more granular basis within buildings, according to the distribution of people within them.

Alternately, Ratti added, existing spaces can be used for new purposes — such as a once rarely used large common area in MIT’s Building 9, which now houses its Steam Café, and thus has chairs and tables occupied throughout the day. “That’s a much more intensive and better use of the space, with the same embedded energy cost,” Ratti said.

Ratti and the Senseable City researchers are currently pursuing several other studies that use IT data to track the flow of people in urban areas, and say they would like to continue to analyze energy use as part of their ongoing work.

Publication date: 04/23/2012