By David Brill

After sunset on August 14, 2003, the residents of New York City who peered upward got a glimpse of something rarely visible in Gotham. Above the darkened city, beyond the towering skyscrapers, spread a canopy of winking stars.

The hundreds of unfortunates pinned in elevators or trapped underground in inert subway cars were afforded no such view, though all New Yorkers could trace their descent into darkness to northeastern Ohio. There, at about 4:00 p.m., a chance encounter between a sagging, overloaded power line and a tree triggered a cascading power outage that affected 50 million people in seven states and Canada. For some, the blackout lasted 30 hours.

Yilu Liu, the UT-ORNL Governor’s Chair for power systems in the College of Engineering, is working to devise a smarter electric grid that automatically resolves minor disruptions before they escalate to major blackouts. She believes part of the answer may lie in a powerful but low-cost technology housed discretely in a small beige box called a Frequency Disturbance Recorder, or FDR.

Liu’s FDR is to the nation’s power grid what an electrocardiogram is to a hospital patient. While the EKG measures the electrical activity of the human heart, FDRs networked via the Internet are able to register the electrical pulse moving through the wires that connect power stations to toasters, air conditioners, and TVs.

Taking the Pulse

Currently, the frequency monitoring network, or FNET, housed in UT’s Power Information Technology Laboratory, tracks 80 FDRs placed around the country. The number will ultimately grow to 2,000 along with the GridEye units deployed by ORNL. At this number, it will provide high density coverage of the U.S. grid.

FDRs are capable of measuring subtle fluctuations in power supply by sampling the frequency, voltage, and phase angle (collectively known as phasor) 1,440 times per second. The unit’s GPS receiver provides a timing signal—precise to the microsecond—as the unit’s algorithm detects and records the fluctuations. All information is sent in real time via the Internet to servers at the Power Information Technology Laboratory, where all data are archived.

Minor disruptions, though detectable by the FDRs, are of little consequence. Not so with major disturbances, like the one that resulted in the blackout of 2003, when the normal 60-hertz frequency dropped as low as 52 hertz in some regions.

Getting the Global Picture

FNET went online in 2004. Had it been fully operational a year earlier, it could have provided power-system operators with the real-time information necessary to avert the worst blackout in U.S. history, or at least to limit its scope and duration.

Instead, system operators relied on data provided by the existing Supervisory Control and Data Acquisition network of monitors mounted at substations. SCADA, though useful, has limitations, particularly when compared to FNET. While FNET takes 1,440 measurements per second, SCADA logs one measurement every 2–4 seconds. FNET measurements are GPS-time-synchronized, while current SCADAs are not. And while SCADA tends to focus on local anomalies, FNET assesses the health of the entire U.S. grid.

“With the tools available in 2003, system operators could not recognize the severity of the situation until hours later,” says Liu. “There was no global picture of the entire system. Everyone was looking at a relatively small piece of the larger puzzle.”

FNET enables dramatic visualization of such disturbances on a map, with the contented green of a system in stasis suddenly flashing in waves of red, orange, blue, and purple, representing wildly fluctuating frequencies. The waves emanate outward from the source of the disturbance.

For example, when nuclear reactors unexpectedly tripped offline in Florida in 2008—resulting in a statewide power outage—FNET’s sophisticated triangulation algorithms processed its units’ GPS-timed signals to instantly locate the source of the disturbance. Today automatic alerts go out instantaneously to the organizations responsible for monitoring and maintaining the flow of electricity through the grid.

If the event threatens a major power disruption, these organizations could then take action to avoid systemic collapse and prevent damage to equipment by strategically shutting down individual loads and power-generating facilities or reducing the voltage supplying the system.

“We may not be able to predict that a certain power line will touch a tree,” says Liu, “but we can detect disturbances that can cause major disruptions.”

Leveraging the Best Resources

Liu’s work represents a critical lobe of what will evolve into the brain of a smart grid, and she credits the wealth of shared resources at UT and ORNL with major advances in her research, as well as the support from EPRI and some 20-plus major electric power utilities and manufacturers.

The project relies on UT’s Kraken and ORNL’s Jaguar supercomputers, among the world’s fastest, to simulate, much more quickly than real time, the potential ripple effects of disruptions to the power supply.

Eventually, data such as those created by FNET will feed into a “self-healing” system of intelligent algorithms capable of triggering automatic remedial action.

“When major disruptions occur, there is no time for human response,” says Liu.

Most Americans take for granted the steady flow of electricity until they find themselves fumbling in the dark for a flashlight. Much worse are the effects of large-area outages on industry, commerce, transportation, and national security. Power outages are troublesome and expensive. According to a 2004 study by Lawrence Berkeley National Laboratory, outages cost the United States about $80 billion a year.

Liu makes it clear that even a smart grid will be susceptible to occasional disruptions, but she and her colleagues are determined to reduce their frequency and range—something that will benefit all Americans.

“We’re all connected to the system,” she says. That’s not necessarily a good thing, if a toppled tree in Ohio can turn out the lights in New York City. Someday, though, as Liu’s research pays off, a descent into darkness will be an entirely voluntary pursuit.

 

Tags: Electrical Engineering and Computer Science • Governor’s Chair • Power Systems • Yilu Liu