Skip to main content

Feature Stories

Power Trip
Model-based Energy Management Systems Drive ‘Smart’ Buildings
A close up view of wires and circuits.

Applying his research methods, Irwin estimates consumers could save 15 to 20 percent on their electricity bills, with similar energy savings for large commercial buildings.

As energy costs continue to rise and squeeze home and large building owners alike, monitoring energy use is becoming as essential as securing alternative energy sources. To address this need, UMass Amherst computer engineer David Irwin draws on his wide-ranging expertise in green computing, embedded sensor systems, and next-generation networks to design and build complex cyber-physical systems (CPSs) that are robust, efficient, and secure.

In one project, Irwin is designing and building software prototypes to manage energy systems to create ‘smart’ buildings. A five-year Faculty Early Career Development (CAREER) grant from the National Science Foundation will lay the foundation for this research as Irwin creates a “Wikipedia-style” website to collect electricity use data from thousands of specific brands and models of appliances. It is designed to be an encyclopedic site where visitors can use data or add data about their appliances. Irwin says understanding how and why individual electrical devices consume electricity is critical to improving a building’s overall energy efficiency. 

Using the site, Irwin will conduct research that uses models of these devices to develop automated discovery, monitoring and scheduling software to identify wasted electricity in a building, track energy consumption, and program electrical devices to go off or on, according to need. This kind of smart electrical system will be inexpensive, private, reliable and sustainable, he says.

David Irwin, electrical and computer engineering
"The purpose of my research is to make a home or building as ‘smart’ as possible in terms of monitoring and controlling energy efficiency,” explains Irwin. “By employing the methods I’m researching, consumers could save an estimated 15 to 20 percent on their electricity bills, while also reducing their carbon footprint.” By extension, that percentage of savings in a large building such as a skyscraper would translate to an enormous amount in money, electricity, and environmental impact.

“It’s not altogether about costs,” Irwin adds. “There’s also a crucial environmental aspect to it.”

As Irwin notes, society now depends on a massive amount of largely “dirty” energy to sustain itself. Collectively, buildings consume significantly more energy – 41 percent of the total amount – than the two other broad sectors of consumption, industry (30 percent) and transportation (29 percent). Typically, electricity is used in homes, apartments, and other buildings to supply a variety of devices, including heating and air conditioning systems, resistive and motor-driven appliances, information technology devices, and lighting. Even small improvements in the collective efficiency of this electrical equipment in buildings of every kind have the potential for a massive impact.

A significant barrier to improving building energy efficiency is that fine-tuned, pervasive monitoring of electrical devices at large scales remains impractical because it is prohibitively expensive, invasive, and unreliable.

Irwin’s research is addressing this barrier in several ways. First, Irwin is creating a “Wikipedia-styled” website to use as a repository for electricity usage data used from thousands of specific brands and models of appliances; an encyclopedic site in which visitors can either use data or add data about their own appliances. Understanding how and why individual electrical devices consume electricity is critical to improving a building’s overall energy efficiency. “Our hypothesis is that a comprehensive repository of detailed models that describe how specific devices use power provides a foundation for programming and managing energy usage in buildings,” Irwin explains.

Having established what in effect is an “Electripedia” of specific electrical devices and the power they consume, Irwin will introduce a new line of research that uses models of these devices to develop automated discovery, monitoring, and scheduling software, which can automatically identify wasted electricity in a building, track energy consumption, and program electrical devices to go off or on, according to need. This kind of smart electrical system will be inexpensive, private, reliable, and sustainable.

Prashant Shenoy, computer science
In related research, Irwin is collaborating with computer scientist Prashant Shenoy to develop a system called Yank, which will enable data centers to "pull the plug" on servers if the level of available power suddenly changes. Yank is designed to seamlessly transition applications to still-active servers until the power is restored. The system allows data centers to better handle power variations from renewable energy sources that can be unreliable.

Large data centers such as the new Massachusetts Green High Performance Computing Center in Holyoke are designed to operate tens of thousands of computers simultaneously. Technology companies such as Google, Apple, Facebook, and Amazon host their Internet services from many data centers spread around the globe. Data centers require an astonishingly large amount of power to operate.  Powering and cooling them represents nearly 50 percent of their total cost of ownership according to Irwin, and recent estimates attribute nearly 3 percent of the entire electricity usage in the United States specifically to data centers. 

The rise in power usage has led data centers to experiment with the design of the power delivery infrastructure, including the use of renewable energy sources. The Yank’s new high availability techniques being designed by Irwin and Shenoy will ensure software services remain available during unexpected power outages or shortages.

Ultimately, society depends on a massive amount of, largely “dirty,” energy to sustain itself.  Thus, meeting society’s future energy needs, while mitigating its economic and environmental impacts, continues to be a critical challenge. The projects described here reflect Irwin’s broader research agenda, which aims to address these challenges by leveraging advances in computation, networking, and embedded sensing.

College of Engineering