Oil in standing water outside Ludden pumping station, TransCanada Keystone pipeline system, May 9, 2011, Sargent County, North Dakota

Oil in standing water outside Ludden pumping station, TransCanada Keystone pipeline system, May 9, 2011, Sargent County, North Dakota (Photo by Pete Carrels)

As crews haul away hundreds of tons of tar-sands-contaminated soil and water from TransCanada's tenth leaky pumping station, I have pointed out that TransCanada assured our state government that such leaks would happen only once every twelve years. Ten leaks in less than one year is at least 120 times the predicted rate of failure, isn't it?

TransCanada now tries to take that risk assessment off the table:

According to the company's risk analysis on file with the South Dakota Public Utilities Commission, a leak of 50 barrels or more on the Keystone system would be expected once every seven years. The pipeline began moving oil in June.

[TransCanada spokesman Terry] Cunha said this estimate doesn't apply to pump stations, which weren't included in the risk analysis. Including Saturday's spill, this is the 10th release of oil on the Keystone line, all of them at pump stations, Cunha said [Cody Winchester, "Valve Failure at Pump Station along Keystone Pipeline Causes Spill," that Sioux Falls paper, 2011.05.10].

Let's review the exact text of the risk analysis in question:

Of the postulated 1.4 spills along the Keystone Pipeline system during a 10-year period, the study's findings suggest that approximately 0.2 would be 50 barrels or less; 0.8 would consist of 50 to 1000 barrels; 0.3 would consist of between 1,000 and 10,000 barrels; and 0.2 would contain more than 10,000 barrels (Appendix A). The spill volume frequency distribution likely underestimates the proportion of spill volumes under 50 barrels due to reliance upon the greater than 50 barrel reporting criteria within the USDOT incident database. The current analysis tends to overemphasize large spills and underreport the small spills, making the assessment conservative.

Based on probabilities generated from the study, the estimated occurrence intervals for a spill of 50 barrels or less occurring anywhere along the entire pipeline system is once every 65 years, a spill between 50 and 1,000 barrels might occur once in 12 years; a spill of 1,000 and 10,000 barrels might occur once in 39 years; and a spill containing more than 10,000 barrels might occur once in 50 years. Applying these statistics to a 1-mile section, the chances of a larger spill (greater than 10,000 barrels) would be less than once every 67,000 years [emphasis mine; ENSR Corporation for TransCanada, "Pipeline Risk Assessment and Environmental Consequence Analysis," Document No. 10623-004, June 2006].

Anywhere. The entire pipeline system.

I've been to Canada. I know their English differs a bit from ours. However, I'm fairly confident that the words anywhere, entire, and system mean the same thing in Alberta as in South Dakota.

Pumping stations are an integral part of the pipeline system. They are certainly somewhere along it. Does TransCanada seriously believe it can absolve itself by claiming that spills at the pumping stations don't count as risk within its pipeline system?

Maybe TransCanada submitted a separate risk assessment on the pumping stations. If they did, I'd be happy to receive and publish a copy. But if no such separate assessment exists, Cunha's statement suggests that TransCanada left the leakiest portion of its project out of its risk analysis.

Update 2011.05.12 10:25 CDT: "The mind reels"—Forbes blogger Osha Gray Davidson agrees that TransCanada's revelation that it didn't include the pumping stations in its risk analysis is a doubt-raising fine-print trick.