By David Sassoon, InsideClimate News, May 30, 2012
Control room technicians 1,500 miles away didn't understand that the 16 high priority alarms that sounded were warning them of a leak.
A ruptured pipeline near Michigan's Kalamazoo River leaked oil for more than 17 hours, even as 16 high-priority alarms sounded in the operator's control room in Canada. Control room workers restarted the pipeline twice during that period—and were preparing for a third restart—when they learned from an outside party about the massive spill in 2010.
Those details were included in documents released last week by the National Transportation Safety Board, which has been investigating the accident for almost two years.
The 30-inch pipeline, called Line 6B, is owned by Enbridge Energy Partners, the U.S. branch of Enbridge Inc., Canada's largest transporter of crude oil. Line 6B spilled more than 1 million gallons of oil into the Kalamazoo River and a nearby creek, making it one of the biggest oil pipeline spills in U.S. history, and the most expensive, with costs surpassing $700 million. More than 30 miles of the river remain closed to the public today, as the cleanup continues. The Michigan accident occurred while oil was still spewing from BP's exploded well in the Gulf of Mexico, and it never received much national attention.
The NTSB, which is expected to reveal the results of its investigation sometime this summer, released hundreds of documents totaling thousands of pages last week. Included in the trove is a factual report compiled by two Enbridge employees and three federal investigators. It provides an hour-by-hour, and sometimes minute-by-minute account of the unfolding disaster as experienced by Enbridge operators, analysts and supervisors who were monitoring computer consoles, hearing the repeated alarms, and discussing what to do in the company's Edmonton, Alberta control room.
The report does not quantify how much oil escaped from the initial rupture, or how much additional oil was pumped out of the 6 1/2-foot tear in the line when it was restarted twice. The NTSB did not respond to questions from InsideClimate News about how much oil may have leaked from Line 6B at these different stages of the unfolding series of events.
The accident began when operators were shutting down 6B as part of a routine procedure to allow the next shipment of oil to accumulate at the start of the line in Griffith, Ind. An alarm sounded and the pump station at Marshall automatically went into a local shutdown within minutes of the rupture, which investigators say occurred at 5:58 pm eastern time on July 25, 2010—a Sunday evening. But at 4 a.m. and again at 7:10 a.m., operators at Enbridge's control room restarted the line. During the first restart, technicians pumped oil under high pressure into the line for one hour; during the second restart, for at least 45 minutes.
Enbridge operators restarted the line because they had concluded that the alarms were being triggered by "column separation," or an air bubble in the line, and that the problem could be corrected by increasing the pumping pressure.
Distinguishing between an air bubble and a leak can be very difficult, said Richard Kuprewicz, the president of Redmond, Wash.-based Accufacts Inc. Though he is not directly involved in the investigation of Enbridge's Pipeline 6B, he has almost 40 years of experience as a pipeline safety expert.
"When your line isn't full of liquid, leak detection is many times more difficult," Kuprewicz said. "It's hard enough to find a rupture when the line is running liquid full."
When there are problems with a pipeline, a leak detection system will often trigger multiple alarms. Ninety-nine times out of 100 the alarms are false, Kuprewicz said. But all of that noise makes it difficult for operators to distinguish a false alarm from an alarm signaling real danger.
Line 6B had experienced three column separation events during the previous 12-months: one in October 2009, a second in April 2010, and a third in June 2010, about one month before the Michigan rupture.
No doubt, Kuprewicz said, NTSB investigators are exploring how Enbridge operators are trained to distinguish a real alarm from a false one. He said he and other safety experts have been prodding the federal government to improve alarm management in control rooms. Beefed up regulations would force pipeline companies to track the type of alarms they receive and how they respond to them.
When asked to comment on the findings in the NTSB document, an Enbridge spokesman directed InsideClimate News to a statement on the company's Web site. "Until the final report is published (by NTSB), we do not intend to pre-empt those findings by commenting on specific details," the statement said. "Enbridge has already implemented, in 2010 and 2011, appropriate operational and procedural changes based on our own internal investigation."
The investigators from PHMSA—the Pipeline and Hazardous Materials Safety Administration—who assisted in compiling the control room report, also declined to comment on the control room decision-making because the investigation is ongoing.
When Enbridge employees finally realized that pressure in 6B had dropped because of a leak, not an air bubble, oil had already blackened Talmadge Creek and was flowing into the Kalamazoo River. The oil was stopped before it reached Lake Michigan, which along with the other Great Lakes supplies drinking water to 26 million Americans.
The cleanup has been difficult because Line 6B was carrying diluted bitumen, or dilbit, an unconventional form of oil derived from Canada's oil sands that has defied traditional oil recovery methods. The long-debated Keystone XL pipeline, which would run from Alberta to the U.S. Gulf Coast if the entire project is approved, would also carry dilbit. The most controversial portion, in Nebraska, would cross one of the largest and most important freshwater aquifers in the nation, the Ogallala. TransCanada, the builder of the Keystone XL, plans to use the same type of leak detection system as Enbridge used on Line 6B, according to TransCanada's website.
In the Enbridge control room, the first six high-priority alarms sounded within five minutes of the initial rupture on line 6B. Two hours later at 8 p.m., the controllers on duty left work at the end of their normal 12-hour shift. According to the documents, they didn't inform their replacements of the six alarms from Marshall, and by the time the new team arrived the alarm indicators had cleared and disappeared from the computer screens in the control room.
At 4 a.m., the controllers on the new shift in Edmonton restarted the pipeline to end what they thought was a planned shutdown. Within minutes, three alarms rang in succession at 4:12, 4:16, and 4:22. Pumps were pushing oil into the 30-inch line with the force of a fire hose. Between 4:36 and 4:57, three more high-priority alarms sounded. At about 5 a.m. line 6B was shut down again. The Enbridge team decided to call it a "false alarm."
The same team tried a second restart at 7:10 a.m. High priority alarms continued to sound at 7:12, 7:35, 7:37, and 7:42. The team called for more pressure but there wasn't enough power available, so the line was shut down at about 8 a.m.
According to the documents, the next shift took over without clear knowledge of the earlier events. Operators were preparing for a third restart when the emergency phone rang in the Edmonton control room. A local gas utility worker in Marshall was on the line with news of the spill. At that point Enbridge initiated its emergency response procedures.
InsideClimate News reporters Elizabeth McGowan and Lisa Song contributed to this report.