ÿ�մ���

UM-Dearborn students, faculty and staff are coming off a win in 2020’s E-Challenge, the annual statewide college energy efficiency pilot program sponsored by DTE Energy. That first-place project was noteworthy primarily for its scope: The 53-member team put together an exhaustive, 500-page, building-by-building energy reduction plan that could save the university $200,000 a year. This year, we’d say the proposed E-Challenge project is apt to score more points for innovation. The main idea: Use the brand new ELB as a building-sized laboratory to design creative, low-cost ways to reduce energy consumption.

Trying to shrink the energy footprint of what’s already our most energy efficient building might initially seem counterintuitive. But Associate Professor of Electrical and Computer Engineering Wencong Su, one of the faculty heading up the E-Challenge team, says that’s part of what makes the ELB an ideal research site. For one, the building’s modern control systems for lighting, ventilation and thermostats give them the opportunity to explore things like AI-based solutions for fine tuning those systems. (That would be way more difficult and expensive in a building with old systems). Plus, there’s already plenty of good research on how to retrofit older buildings for better performance. How to squeeze more savings out of an already high-performing building — that’s way more of an unsolved problem. It’s also an increasingly important question now that many institutions and companies are burning through the list of low-hanging-fruit efficiency measures, like LED lighting, in a quest to get to zero carbon emissions.

The UM-Dearborn team is proposing to tackle this challenge by combining two distinct approaches. One looks to smart sensors and artificial intelligence, the other to some insights from behavioral economics. But both approaches depend on solving an initial data problem. Su says even in new buildings, you typically won’t have granular data that reveals how much energy a particular plug, light fixture, lab instrument, individual lab or classroom is using. Yet such information is crucial for creating AI-based control systems, or for coming up with non-AI solutions for that matter. So as a first step, the E-Challenge team will have to figure out how to capture this more detailed energy profile of the building — and do it cheaply and in ways that preserve privacy. For example, Su says part of their system will likely involve deploying occupancy sensors to measure the density of use in particular areas of the ELB. Camera-based sensors can be great for that, but because they also capture identifiable faces, they can feel a little Big Brother. Conversely, infrared sensors can be great for protecting identities, but they also have to be able to distinguish between the heat profile of a human being and that of a computer or lab instrument. So getting the tech right  — at the right price — will provide an key initial challenge for the team.

Another big challenge: How to deploy a network of sensors that doesn’t compromise the cybersecurity of the building. “In older buildings, everything tends to be hardwired, but with modern self-powered sensors, your connectivity is usually a wireless communications network,” explains Assistant Professor of Electrical and Computer Engineering Junho Hong. “That potentially introduces new vulnerabilities, and you wouldn’t want someone hacking into your HVAC system and shutting down the air conditioning in the summer.” Hong says issues like this are actually a really hot topic in the cybersecurity industry right now, so it’s a great opportunity for students to get experience with an emerging real-world problem.

If they can successfully navigate those challenges, the result will be a ton of useful room-level data that could be the basis for all kinds of solutions. AI-powered tweaks to building management systems or class scheduling that optimizes building energy use are a couple ideas the team will be investigating. But Assistant Professor of Economics Antonios Koumpias says they also plan to use that data to try to inspire behavior changes. For example, research indicates that simply knowing how much energy you’re consuming can have an effect on your usage. This is the idea behind those postcards you sometimes get that tell you how your neighbors’ energy bills compare to your own; or the notifications you get from your smart thermostat app showing how your energy use ranks for your zip code. And the team could even leverage that data to create a more formal department- or lab-level competition to see which units can have the smallest per capita footprint. For example, Koumpias says publishing that data in a public place, like the ELB’s new 29-foot big screen, could have the effect of rewarding good behavior and nudging bigger energy consumers into adopting different habits. Both positive behavior rewards and forms of public “shaming,” he says, have been shown to be powerful motivators in other domains. (For example, Koumpias says some states use these methods to discourage tax delinquency.) So he’s curious to study how nudges, competitions and shaming could work for collective energy conservation.

In terms of project design, however, Koumpias says one of the biggest challenges will be getting buy-in from people using the ELB. “Nudging individuals is a little more straightforward, but nudging groups of individuals tends to be more difficult,” he explains. “For example, for a group of students in a lab to shift their behavior, they need to feel like there is a community between them, so it’s worth it for them to do it together. If the groups are not socially cohesive, it’s unlikely that you’ll change the behavior.” In other words, if they don’t feel like a team, they likely won’t care much about competing as a team and beating out the other teams. 

Though this year’s E-Challenge project will focus specifically on reducing the energy footprint of the ELB, the team hopes their work could lead to low-cost and no-cost solutions that could be deployed in other modern buildings. They’ll spend the next two months refining their proposal, and then, pending approval, they’ll have until April 2023 to run their building-level experiment. 

###

Story by Lou Blouin. This year’s E-Challenge team is being led by UM-Dearborn faculty Wencong Su, Antonios Koumpias, Natalie Sampson, Yi-Su Chen, Mengqi Wang and Junho Hong.