Power failures aren’t that common in the U.S., but they are more likely to happen in the summer, when people crank up the air conditioning. Back in August 2003 a blackout covering much of the northeast happened because the control systems for the grid system did exactly what they were designed to do — but they didn’t account for a combination of failures that was unusual. A similar thing happened in 2012 in India. Small failures can cascade into big ones.
To address this problem, a group of computer scientists at MIT wrote a computer program to help engineers find the kinds of glitches that could darken whole cities. Failures tend to occur in pairs — one feeding into the other. So the software identifies the most dangerous pairs of failures among the millions of possibilities and further narrows those to the ones most likely to cause damage. It’s a monumental sifting task, but computers are better equipped for it than people.
Konstantin Turitsyn, the an assistant professor in MIT’s Department of Mechanical Engineering and graduate student Petr Kaplunovich wrote the program, and will present it at the IEEE Power and Energy Society Meeting in July.
To see how well it works, the scientists tested the software on data from a power grid in Poland consisting of 3,000 components and up to 10 million potential pairs of failure. Within 10 minutes, the algorithm weeded out 99 percent of failures, deeming them relatively safe. The rest were pairs that would turn into major blackouts if left alone. The calculations could be done on ordinary laptops.
Turitsyn told MIT News that he plans to test the algorithm on bigger models — like the one that supplies the northeastern United States. In that case it will deal with a system that consists of more than 100,000 components and five billion potential pairs of failures.
If power companies can implement something like this power companies would have a chance to deal with problems while they are still small, and anticipate where failures might happen. At the very least they could get the workers and equipment ready to go near places where they are more likely to be needed.
via MIT News
Credit: Wikimedia Commons / Camerafiend