Technology key to anticipating, preventing outages

Friday, August 22, 2003

NEW YORK -- Every two to 10 seconds, the power grid operator for a group of Mid-Atlantic states collects thousands of measurements on voltage, circuit breaker positions and the like.

Computers at PJM Interconnection summarize the information in graphic form and run what-if scenarios every five to 15 minutes to project whether a hypothetical failure in one component could be catastrophic.

PJM's equipment, installed just two years ago to do calculations more frequently and thoroughly, is among the more modern monitoring technology on an otherwise rather backward national power grid, where engineers rely primarily on telephones to alert neighboring regional power pools to trouble.

In last week's massive blackout, the telephone warnings for much of the system came too late or not at all.

Only a small part of PJM's coverage area was affected.

It's too early to say whether the systems at PJM, which operates the energy market from New Jersey to West Virginia, were decisive in containing the outage, but the commitment to technology is a sign of where the electricity industry is heading in trying to prevent cascading outages.

Computer systems could soon get smart enough to anticipate a growing tree branch about to rupture a power line, and sensors could detect an insulator about to break, allowing crews to make repairs before equipment overheats, experts say.

"Our vision of the grid of the future is what we are calling the self-healing grid," said Luther Dow, director of power delivery and markets at the industry-funded Electric Power Research Institute in Palo Alto, Calif. "It's one that monitors itself, measures itself and even takes corrective action to eliminate reliability problems."

A self-healing grid is still a decade away, though Dow said technology to anticipate failure is already being tested by some 10 utilities.

IBM Corp. and upstarts like SmartSignal Corp. are working on smarter ways to analyze data, taking advantage of new sensors that communicate wirelessly to permit quicker, cheaper and more extensive information flow.

"There are so many decisions that need to be made so quickly and so many minute pieces of data, it's too many for the mind to comprehend," said David Samuel, general manager for IBM Energy & Utilities Industry in Atlanta. "This is about programming the computer to do a lot of the analysis."

Some improvements have already been made. Instead of presenting raw data in tabular columns that humans must digest, for example, newer systems analyze and summarize the information in rich, graphical displays, said Mike Unum, a manager for automation and network services at GE Power Systems in Melbourne, Fla.

But more can still be done, Unum said.

Citing the pending investigations, utility officials have said little about what role monitoring technology may have played in the blackout.

FirstEnergy Corp., the Ohio utility at the center of probes into the blackout, has acknowledged that at least one component of its computer monitoring system wasn't working, preventing an alarm from sounding when troubles began.

At the Midwest Independent Transmission System Operator, which manages electricity flow in the Midwest, alarms sound several times daily and engineers are trained to sort through them, spokeswoman Mary Lynn Webster said.

She would not comment on suggestions by some industry analysts that information overload -- or the lack of an efficient way of processing it -- might have been a factor.

Analysts say that when multiple failures occur, control-room operators can spend precious minutes sorting through alarms to figure out what needs attention first.

"It could alert you to too many problems," said Joseph Bucciero, a senior vice president at energy consultants Kema Inc.

PJM's general manager, Robert Hinkel, sees shortfalls with even its newer systems. He wants more automated sharing among neighboring utilities to prevent local trouble from snowballing again and better artificial intelligence that can forecast radical shifts in power loads.

Advanced computing systems that Arizona Public Service Co. has been testing can help operators see the big picture rather than leaving them simultaneously watching separate displays, said Tom Glock, the company's manager of power operations. Such systems represent the future, he said, though they are still too complex.

Another hurdle is money. A self-healing network could cost tens of billions of dollars and require retrofitting old equipment.

Even computer systems deployed today use designs that "date back a couple of decades," said Dave Pacyna, chief executive of Siemens Power Transmission and Distribution Inc., a Raleigh, N.C.-based vendor of monitoring systems for utilities.

"Maybe," he said, "this blackout is an early warning to get moving a little faster."

Respond to this story

Posting a comment requires free registration: