Getting to the Bottom of the Aug. 14 Blackout

Engineers and politicians have been working diligently to discover the cause of the Aug. 14 blackout, which left 50 million people without power in the northeastern United States and Canada. They've pointed fingers at the country's aging transmission system, poor intergrid communications, and the failure of energy deregulation. The simple explanation is that there wasn't enough power available after some power plants and transmission lines tripped offline in Ohio and Michigan. Technically speaking, this is true, but it also raises the question of the type of power involved.

Reactive power and voltage sags.

The power flowing down transmission lines has two components: active power, which is measured in watts, and reactive power, which is measured in VARs. Most references to power concern the former. This is what most individuals pay for in electric bills, and it's what does the useful work derived from electricity.

It's apparent that at some point while the system was collapsing on Aug. 14, not enough active power was getting to the loads to keep the motors spinning. However, the real culprit may prove to be the lack of reactive power. According to a Sept. 26, 2003, New York Times article:

“Experts now think that on Aug. 14, northern Ohio had a severe shortage of reactive power, which ultimately caused the power plant and transmission line failures that set the blackout in motion. Demand for reactive power was unusually high because of a large volume of long-distance transmissions streaming through Ohio to areas, including Canada, than was needed to import power to meet local demand. But the supply of reactive power was low because some plants were out of service and, possibly, because other plants were not producing enough of it.”

Reactive power is necessary to maintain the voltage to deliver active power through transmission lines. Motor loads and other loads require reactive power to convert the flow of electrons into useful work. When there isn't enough reactive power, the voltage sags and it isn't possible to push the power demanded by loads through the lines.

A major power quality characteristic of the Aug. 14 blackout was sagging voltage in portions of the Ohio and Michigan transmission systems that began at about 3 p.m. This followed a loss of both a generating plant in northern Ohio and a transmission line that feeds Cleveland. By 3:30 p.m., a second line that feeds Cleveland went down, causing voltage to drop even more. Fifteen minutes later, northern Ohio lost two additional lines and Cleveland experienced severe low voltage. At 4:09 p.m., northern Ohio became isolated from the grid and voltage in Michigan weakened. By 4:10 p.m., within a 9-sec period, the entire eight state region plus parts of Canada collapsed.

Reactive power and VAR demand.

Returning to the issue of reactive power, the New York Times article indicates that there wasn't enough VAR supply to meet demand after some transmission lines and generating plants tripped offline. Part of the reason for the high VAR demand was that power was traveling long distances to meet contractual obligations in the energy markets.

Inadequate VAR supply leads to low voltages on the transmission system. As the voltage drops, the current must increase proportionately to maintain the supply of power, which causes the voltage drop to increase. In a reasonably strong system, this process eventually reaches equilibrium and the worst outcome is that losses increase and system efficiency suffers. However, the grid on Aug. 14 had become very weak, so this cycle of decreasing voltage and increasing current eventually went into a runaway condition, yielding what power engineers call “voltage collapse.” Once this happens, automatic controls kick in to attempt to sectionalize the system into stable islands. After a time delay, generating units will shut down to protect themselves if they perceive that the system is unstable. Ties between utilities will break to contain the problem.

Power monitoring data.

Power monitoring data captured at numerous locations across the affected area was useful in understanding the full cascade of events, notably the escalating series of voltage sags, as shown in the accompanying recorded plots. In fact, customers who used monitoring systems could recognize the unique signature of the simultaneous 3-phase deep sag and were able to affect an orderly transition to their UPS and backup systems.

At about 3:45 p.m., a fault apparently occurred on the transmission system. The 8-cycle voltage sag, shown in Fig. 1, was measured at an industrial site in Cleveland. One explanation for why this fault occurred is that the lines overheated from excessive current brought on by load transfers, which resulted from preceding line and generator tripping. Overheated lines elongate and can sag into trees or other structures and lead to a short-circuit fault. The fault was detected and cleared promptly, but the voltage recovery at the site appears to have been slow, suggesting that the system was now much weaker. At least one more line was out of service.

Shortly after 4 p.m., another instantaneous voltage sag was recorded, and the plot shows that the voltage drops abruptly and remains at the lower level (Fig. 2). Phase unbalance developed, suggesting either the presence of a remote fault or that the tripping of another line had caused the system to become very weak.

A short time later, the event shown in Fig. 3 was recorded in an office building in downtown Manhattan. The waveform shows the voltage that triggered the monitor. This waveform is consistent with that of a power system that has become unstable, and the synchronous generators are “slipping poles” relative to a more distant part of the system that remains in synchronism. Once this occurred, the power system that supplies Manhattan immediately went down and separated from surrounding power systems, several of which remained viable throughout the blackout period.

Once the massive amount of load in the affected areas was lost, the entire Eastern Interconnection experienced a jump in frequency of about 0.2 Hz that was seen over a large geographic area. After a few minutes, generator controls brought the average frequency back to 60 Hz and few energy users outside the affected area realized that anything had happened. While large in terms of system dynamics issues, this frequency change is inconsequential to most loads.

Power quality issues also arose as the utilities began to restore power piece-by-piece. In this situation, there typically will be several switching operations and moments when there is a large mismatch between load and available power. This mismatch usually results in either high or low voltages. In fact, the system might get into resonance conditions that it wouldn't normally experience. Fig. 4 shows one of a series of voltage sags that occurred during the restoration at the monitor site.

For most of the homes and businesses affected by the blackout, the lights came back on the next day, but the damage had been done. Early numbers from New York estimated that the blackout cost the city $500 million in lost revenue, making it an expensive lesson in the proper management of power distribution and a powerful reminder that reactive power can be just as expensive as its active counterpart.

Dugan is a senior consultant with Electrotek Concepts in Knoxville, Tenn., and Leinfuss is vice-president of marketing with WPT/Dranetz-BMI in Edison, N.J.