|
BOEM study analyzes North Slope spills Finds that the number of spills relates to oil production rates but that large spills tend to be rare and somewhat unpredictable Alan Bailey Petroleum News
The Bureau of Ocean Energy Management, or BOEM, has published the results of a study that the agency commissioned into the frequency and nature of oil spills from the Alaska North Slope oil infrastructure. Carried out by Nuna Research and Planning Group LLC, the study found that spills of oil volumes greater than 500 barrels are rare, random events. Small spills, on the other hand, appear to occur at frequencies that relate to oil production rates and tend to follow a somewhat predictable annual cycle.
But the dominance of just a few spills, much larger than the rest, obscures any statistical relationship between spilled volumes of oil and factors such as oil production.
1,577 spills The study obtained data dating from 1971 to September 2011 from seven datasets sourced by BOEM, the Alaska Department of Environmental Conservation, the National Response Center and Nuka Research. The analysts only considered spills of one barrel of oil or more and did not include spills associated with the trans-Alaska oil pipeline. That resulted in a total count of 1,577 spills. Of those spills, 1,300 were of 10 barrels or less; 250 were between 10 and 200 barrels; and just two were greater than 1,000 barrels.
The largest spill, amounting to 5,053 barrels, was the 2006 spill at Gathering Center 2 in the Prudhoe Bay field. That one spill accounted for 15.5 percent of the total oil spilled on the North Slope, the study reported.
Mechanical failure An analysis of the reasons for the various oil spills revealed some form of mechanical failure in the infrastructure to be the most common cause. Mechanical failure as a result of corrosion factored high in terms of volumes of oil spilled, a result strongly influenced by that Gathering Center 2 incident. Over-pressurization of equipment came second to corrosion in its contribution to the total spilled volumes.
Part of the purpose of the study was to conduct a statistical analysis of the spills, to try to identify parameters that might be used as predictors for the possible future occurrence of spills of various sizes. For this component of the study, the analysts decided to only use data from the 1,492 spills that happened between Jan. 1, 1980, and Dec. 31, 2010, thus eliminating some unreliable data prior to 1980 and avoiding the use of data from a partial year in 2011.
Pattern for small spills Plots and statistical modeling of this data revealed that the frequency of small spills, especially spills of less than 10 barrels, followed a cyclical pattern during the course of each year, peaking in June and dropping to a minimum in October — the study report does not offer an explanation for this pattern. On an annual basis, the number of small spills tended to correlate with annual oil production, dropping over time as oil production declined. However, there was a break in this pattern in the early 1990s, at which time relatively few spills were recorded.
Larger spills do not exhibit any particular trend over time. Similarly, the study found that the total volume of oil spilled each year does not follow any evident pattern: Spills ranging in size from 10 barrels to 200 barrels accounted for the bulk of the volume spilled, except for the 2006 Gathering Center 1 spill, which generated a large spike in the profile of annual volumes spilled.
Correlation with production A statistical analysis confirmed the correlation between the number of spills each year and the annual oil production from the North Slope, with low spill frequencies in the years 1992 to 1995 appearing as an unexplained anomaly. An analysis which split out separate data for the Prudhoe Bay, Kuparuk River and Milne Point fields indicated that the correlation between oil production and spill frequency is consistent across the fields, with differences in spill frequencies between the fields being accounted for by factors such as the total length of pipeline in operation in each field and differences in field oil production volumes.
The analysts, unable to identify any parameters that could be used as predictors of oil spill volumes, recommended making volumetric predictions by first estimating the numbers of spills likely to occur for different ranges of spill size, and then multiplying the number estimate for each range by the average volume of oil spilled within that range.
Few large spills But the analysts found that large spills in excess of 500 barrels are rare and random, with an analysis of the historic records resulting in a prediction of zero to two spills of this magnitude for every billion barrels of oil produced on the North Slope. And, with loss of integrity of some part of the oil infrastructure being the most common cause of an oil spill, examining the spill records of individual facilities within the infrastructure and hence establishing spill occurrence rates for different facility types may provide additional insights into spill occurrence rates, the study report says.
|
