An investigation of pipeline accident reports from the last ten years has revealed that the much touted leak detection systems employed by pipeline companies only catch one out of twenty spills. The InsideClimate New article by Lisa Song illustrates an alarming disconnect between industry rhetoric and reality when it comes to detecting leaks on pipelines. Not only do pipeline leak detection systems miss nineteen out of twenty spills, they miss four out of five spills larger than 42,000 gallons. Understanding the limits of current leak detection technology has never been more important. As companies like Enbridge and TransCanada propose pipelines moving large volumes of tar sands across sparely populated areas, through rivers and aquifers, it’s critical that the public consider what’s at stake with open eyes. Particularly after learning from Enbridge’s Kalamazoo tar sands pipeline spill how much more damaging tar sands can be.
TransCanada has told regulators that its leak detection system has a threshold of between 1.5% and 2%. Given that Keystone XL has a maximum capacity of 830,000 barrels of tar sands per day, TransCanada is saying that Keystone XL’s leak detection system can only reliably identify leaks if they’re spilling more than 500,000 to 700,000 gallons of tar sands a day. When put in that context, the reason folks don’t want Keystone XL built through their rivers and groundwater become clear.
Of course, TransCanada has told federal regulators that “computer based, non real-time, accumulated gain/loss volume trending would assist in identifying seepage releases below the 1.5 to 2 percent” threshold. In plain English, that means that given enough time, if TransCanada put a certain amount of tar sands in one end of Keystone XL, and gets less oil out of another, eventually they’ll determine they have a leak. But when?
Few would take heart upon learning the answer to that question. One of the “57 special conditions” that Keystone XL proponents claim will make the pipeline safer lays out the requirements its “non real time” leak detection system. Condition 31 says that Keystone XL’s leak detection system must be prepared using guidance provided in the Canadian Standards Association (CSA). And what does the CSA say?
To comply with this “special condition,” TransCanada’s non-real time leak detection system must be able to detect spills of 4.9 million gallons within a week (or 2% of its capacity). Leaks larger than 350,000 gallons a day, or 1% of its capacity, must be identified within a month – allowing a leak to generate a spill of over 10 million gallons over the course of a month before discovery. And there is no guidance for leaks less than one percent – on Keystone XL, a leak less than 350,000 gallons a day. When looking into it at way, the condition doesn’t seem that special.
These issues are also at play with Enbridge’s Northern Gateway project, a pipeline to move tar sands across the mountains and rivers of British Columbia. As we noted in our report, that 525,000 bpd tar sands pipeline could also leak millions of gallons of tar sands in highly remote regions without its leak detection system identifying a problem.
Enbridge’s Kalamazoo tar sands spill presents another case undercutting industry’s claims about pipeline safety and leak detection. As the InsideClimate piece notes, “Just 10 days before the accident, Enbridge Inc., which operates the Michigan pipeline, told federal regulators it could remotely detect and shut down a rupture in eight minutes. But when the line burst open, it took Enbridge 17 hours to confirm the spill.”
What is more surprising is that one month after failures in its leak detection system allowed it’s line 6B pipeline to spill over a million gallons of tar sands into the Kalamazoo River, Enbridge proposed to employ a new leak detection system only capable of detecting leaks greater than 15% of Line 6B’s capacity. Such a leak detection system could only identify spills greater than 1.2 million gallons a day.
While Enbridge is now well known for its “Keystone Kop” performance during devastating Kalamazoo tar sands spill in Michigan, a smaller spill on another Enbridge pipeline demonstrates an entirely different category of risk. In June of 2011, a landowner discovered a 63,000 gallon spill from a leak the size of a pin-hole. No one is clear how long the leak had been ongoing, but one thing is clear – if a landowner had not happened upon the spill, in all probability the pipeline would still be leaking.
Operators can feel pressured to "tell people things they shouldn't tell them because it's not true" Richard Kuprewicz, President of Accufacts, Sept. 19, 2012
This is quite different from the picture painted by pipeline company representatives. In one public panel, TransCanada representatives simply denied that spills smaller than 2% could not be reliably detected by Keystone XL’s real time leak detection system. Simply stated, it’s hard to have an honest public discussion about the risks of projects like Keystone XL when the company sponsoring the project isn’t honest to the public about those risks.
Photo of Kalamazoo River cleanup, courtesy of Mic Stolz