On May 25, 2006, there was a power outage along Amtrak's Northeast Corridor line, a heavily traveled route that stranded over a hundred trains from Amtrak, NJ Transit and other transit companies. Now, nine months later, Amtrak has revealed what went wrong - and it's a doozy. The NY Times reports:
A 4-year-old computer in Philadelphia failed to execute a single command given 36 hours earlier, after maintenance had been done on the evening of May 23, and then failed to alert human controllers that it had not followed instructions, according to an extensive investigation performed by outside experts. The effect was to limit the amount of power available in the system, leaving no margin during periods of peak demand.
The system muddled through the morning and evening rush hours on May 24, but on the morning of the 25th it became overloaded and failed, according to Amtrak. Giant machines, needed to convert power to the type Amtrak needs, failed one by one, first at a substation in Sunnyside, Queens; then at Jericho Park, near Bowie, Md.; and then Lamokin, near the Pennsylvania-Delaware border.
Augh! It all started in Queens, which might as well be known as the center of all power-related problems! The other hilarious thing is that apparently the vendor who set up the power-reducing measure during maintenance couldn't explain why the power-reducing measure was necessary. Or is it funny that the computer that failed was one of two, but the other one was not working because of other problems?
Amtrak has enacted new procedures in the wake of the power outage: There are technicians on site at various sites during rush hours, plus there's rescue diesel train to get stuck trains out of tunnels - last year, a couple trains were stuck in the tunnel between Penn Station and NJ.
Photograph of passengers waiting and being led out of trains last May during the Amtrak power outage from the AP