(The following article by Matthew L. Wald was posted on the New York Times website on May 26.)
WASHINGTON — Amtrak runs an electrical system that in some ways resembles a utility company’s, and on Thursday it had a blackout. The sequence of events that crippled train service from here to New York is still emerging, and what started it, and why it spread so far, are not yet clear.
Technicians have made a list of the circuit breakers that opened, eventually cutting so many sources to the electric power lines over the tracks that the remaining sources, in a cascade, shut themselves down to avoid equipment damage. But at Amtrak’s headquarters here, executives could not explain, even after the system was restarted, what caused the first circuit breaker to trip.
“It could be anything,” said William Crosbie, the vice president for operations. Parts of the system date to the 1930’s, while other parts were installed as recently as this decade, he said. Animals that come into contact with wires can cause short circuits, he said, although the system should weather single failures.
Such failures occur often, but they do not lead to wider ones, he said. But the protective system worked as intended on Thursday, according to Amtrak, preventing any damage to the equipment. And despite massive inconvenience, nobody was hurt.
When engineers establish where the problem started and why it spread, they will probably also answer the question of how the breakdown relates to the generally antiquated state of Amtrak’s Northeast Corridor between here and Boston. The inspector general of the federal Transportation Department reported in late 2004, “Continued deferral brings Amtrak closer to a major point of failure on the system, but no one knows where or when such a failure will occur.”
The inspector general at the time, Ken Mead, said that the railroad was trying to maintain service without investing enough in hardware, which “resulted in a form of Russian roulette, spreading capital much too thinly and substantially increasing the amount of deferred investment.”
Amtrak does not generate its own electricity, but buys it from local utilities. It uses substations, like those operated by Con Ed and other power companies, to adjust that power to the proper voltage. But to run its trains between New York and Washington, it must perform an unusual conversion, changing the frequency of the power — that is, how fast the alternating current alternates — to 25 cycles. The standard frequency in the United States, used to run everything from light bulbs to hair dryers to air conditioners, is 60 cycles.
The sequence of events that follow is still preliminary.
At 7:55 a.m., two heavy-duty circuit breakers on the 25-cycle system at Amtrak’s Jericho substation, near Bowie, Md., on its Northeast Corridor, sensed a problem, and cut the electricity to two circuits that power the overhead lines.
Almost simultaneously, two breakers tripped at a substation in Sunnyside, Queens, about 200 miles north. But the network, with various redundant parts and cross-connections, kept working.
At 8:02 a.m., three more breakers tripped at a substation called Richmond, near Philadelphia. But power to the overhead lines continued.
A few seconds later, as in a game where players take turns removing parts of a tower, one more circuit breaker opened at Richmond and power to the system from here to New York — four parallel tracks running 225 miles — tottered. Operators, seeing huge fluctuations, quickly intervened, not to save the power, but to kill it, and prevent overloads that would fry components that could take months to fix or replace.
The cause of the first problem, at Jericho, is not certain, Mr. Crosbie said. Engineers will have to go to each circuit breaker and download its memory to find out what caused the breaker to activate, and when.
But another factor may be that a substation at Metuchen, N.J., was shut for scheduled maintenance. That left only two others functioning: Safe Harbor, near Lancaster, Pa., on Amtrak’s Harrisburg line near its junction with the Northeast Corridor, and Lamokin, near the Pennsylvania-Delaware border.
It took two attempts to get the system running again; the first, undertaken a few minutes after 9 a.m., resulted in a second crash, as operators mismatched supply with the demand from scores of stranded trains, which began sucking current back into their motors. A second effort was successfully completed by about 10 a.m.
While the loss of power for the motors was total, Amtrak and its thousands of stranded rail passengers were lucky. Power to the signal system was not interrupted, so it was safe to use diesel locomotives to push trains to nearby stations. Power to switches on the rails worked, too.