The freight train derailment last week in Paulsboro likely was the result of human mistakes (overriding a red stop signal), money-saving automation (replacing a human bridge operator with an electronic system), and overloaded old infrastructure (a bridge with parts dating to 1873), rail experts say.
The bridge failure and derailment Friday dumped four tank cars into the Mantua Creek, one of which ruptured, spewing hazardous vinyl chloride gas into the air. Seventy people went to hospitals, and more than 100 residents are expected to remain out of their homes this week while crews try to remove the dangerous chemical.
No one is believed to have been seriously injured.
A critical failing, rail experts said, was the decision by a Conrail dispatcher in Camden to send the train across the bridge even though the bridge signal would not turn from red to green and even though the last train over the bridge reported that the span apparently had not automatically reopened, as it was supposed to do.
The swing bridge had an intricate locking mechanism to align the bridge tracks with those on shore and to secure the bridge to its abutment. All four locks had to be in place for electronic sensors to trigger a green signal.
The engineer of the Conrail train Friday used an electronic key pad, similar to a garage-door opener, to try to get the appropriate green signal, but could not. After the train conductor walked the bridge and reported that it looked secure, the engineer asked for and received permission from the Camden dispatcher to proceed through the red signal.
"It's almost unheard of to run a red signal," said a former Conrail bridge engineer.
Kevin Thompson, spokesman for the Federal Railroad Administration, said a dispatcher may authorize an engineer to operate through a red signal under certain circumstances, "the most common reasons being a track circuit malfunction or a signal malfunction."
The engineer on the train that derailed had been working the South Jersey route for 14 months, and the conductor was on his first week on the route. Both had made uneventful trips over the bridge the previous three nights.
Conrail officials knew there were problems with the Paulsboro bridge mechanism and signal.
Conrail had received 23 "trouble tickets" from crews or others about the bridge in the past year, including nine such reports since Oct. 27, officials of the National Transportation Safety Board said Monday.
A crew report on Nov. 19 indicated that the bridge had not locked properly, and on Thursday, a crew reported the rails were four inches away from being locked, the NTSB said.
Conrail made adjustments to the bridge mechanism on Thursday, and four trains proceeded safely across the span after that, the NTSB said.
The last crew across, about eight hours before the derailment, reported that it had received a "bridge failed to operate" recording after crossing the bridge, indicating the bridge had not automatically opened after the train crossed, as it was supposed to do.
Nonetheless, the dispatcher permitted the next train to proceed against the red signal, and that train derailed about 7 a.m. Friday.
A human operator formerly opened and closed the bridge, but he was replaced in recent years by the automated electronic system.
The automation probably saved Conrail about $100,000 a year, a former employee estimated.
But the move also drew objections from some employees, who said they warned that a human operator was more likely to spot and correct misalignments or unlocked rails.
Conrail spokesman Michael Hotra would not answer questions about the bridge operations or its recent history, saying only that "we are cooperating fully with the NTSB investigation."
The swing bridge was originally built in 1873, and some of the original apparatus remains, although much was replaced after a 2009 derailment that buckled part of the bridge.
The swing bridge's mechanical equipment was supposed to be inspected every three months, but Conrail apparently failed to make the scheduled quarterly inspection in September, NTSB officials said. Conrail declined to answer specific questions about its bridge inspections.
Some rail experts also questioned whether bridge pilings, or underwater supports, might have weakened, especially after Hurricane Sandy. Conrail crews did "cursory, visual" inspections Oct. 30 and 31, right after the storm, NTSB chairwoman Deborah Hersman said.
NTSB officials said underwater inspections were supposed to be conducted every five years, and the next was set for September 2014.
The bridge was built for lighter cars than modern tank cars, although old steam locomotives were very heavy. The heavy use of the bridge over the decades, coupled with its complex swing mechanism, could have contributed to its failure.
"On our most-wanted list of transportation safety improvements is the integrity of the infrastructure in our country," Hersman said in an interview. "We will be collecting information on that while we're here."
Hersman said the NTSB would not determine a probable cause of the accident while its investigators were on the scene.
"We wouldn't be here if something hadn't gone wrong," she said. "We are working to get to the bottom of it, to understand what happened and why it happened, so we can make recommendations to prevent it from happening again."