Spotlight on Research: ConEd and Columbia Work to Prevent Power Outages

Spotlight on Research: ConEd and Columbia Work to Prevent Power Outages

posted 06/01/05

Spotlight on Research: ConEd and Columbia Work to Prevent Power Outages

These images dramatically illustrate the extent of the blackout on August 14, 2003. Scientists at The Earth Institute are working with Con Edison to learn why and how such energy failures occur. Image Credit: NASA.

What can one of the world's most advanced systems of electricity delivery teach other growing urban power grids about reliability, performance and responsiveness to increasing demand? According to Roger Anderson, a Doherty Senior Scholar with the Lamont-Doherty Earth Observatory unit of Columbia's Earth Institute and a specialist in energy management systems, plenty.

Anderson and a team of Columbia scientists have set up a "Columbia" workstation in the Manhattan Electric Control Center (MECC) of New York City's Con Edison, the transmission and distribution company that provides New Yorkers with 10,000 megawatts of electricity daily. The Center is connected to an umbilical system that sends power around Manhattan live, recording at any given moment where electricity has been interrupted and what the expected time delay is before power returns to normal function. But not to worry, redundancy in the electricity network keeps the lights on even when some of these electricity "Feeders" go out.

The Columbia team, which includes Anderson, senior staff associate Albert Boulanger and scientists from Columbia's Center for Computational Learning Systems (CCLS) in the Engineering School, led by David Waltz, are part of a research initiative to introduce machine learning and other "lean" energy technologies developed at Columbia's LDEO and CCLS to the Con Edison system.

As Anderson explains in a recent article on lean energy management, "Machine learning methods effectively combine many sources of information to derive predictions of future outcomes from past performance. Individually, each source may only be weakly associated with something that we want to predict, but by combining attributes, we can create a strong aggregate predictor." By understanding why and how energy failures occurred, researchers can use the same methodologies to prevent future 'train wrecks' — such as the 2003 blackout in the Northeast.

Analyzing vast amounts of data about feeder failures for the past 9 months — some 400 MB's of new data each day — the team is looking for patterns in the data and developing new machine learning methodologies to use in real time to predict when and where feeders in the distribution system are susceptible to future failure. So far, the Columbia team has developed algorithms that predict 40 percent of future outages for the data they have reviewed. Their goal: to be right 80 percent of the time.

The models being developed take into account a range of variables that affect the functioning of the electricity feeder systems — including heat, moisture, the presence of salt (an aggravation during winter months), and the age and condition of the feeder lines. The team was working with historical data until now, but they have just taken their methodology 'live,' moving from static to real-time, dynamic information coming in on the present condition of the electric grid in Manhattan. The Columbia team is developing feedback loops that will help Con Edison's engineers manage operations even more efficiently.

Why this push to computer-based lean energy management? Con Edison, Anderson notes, anticipates that its customer base will swell to 30 million people by 2020, double the current usage. To deliver double the energy, Con Edison's system is going to have to be smarter, more flexible and more proactive in its systems maintenance. The company also will need to know the expected survivability of its existing system — the likely lifetime of existing feeders — so that it can plan over the long-term to replace underground cables.

Lessons learned in one of the most advanced electric grids have broad implications for electric energy development in the developing world. "The information we are gathering on the Con Edison project will help quickly urbanizing areas around the world determine how best to maximize the design of and investments in electric grid infrastructure," says Anderson. "The innovation growing out of this project with Con Edison, which serves one of the most demanding electricity markets, can provide new insight for countries hoping to leapfrog current practice in favor of more reliable maintenance strategies and technologies."

The Earth Institute at Columbia University is the world's leading academic center for the integrated study of Earth, its environment and society. The Earth Institute builds upon excellence in the core disciplines — earth sciences, biological sciences, engineering sciences, social sciences and health sciences — and stresses cross-disciplinary approaches to complex problems. Through research, training and global partnerships, it mobilizes science and technology to advance sustainable development, while placing special emphasis on the needs of the world's poor. For more information, visit