A new paper released from a graduate student and associate professor at Stony Brook University in New York suggests that Marcellus development “poses substantial potential risks of river and other water pollution.”

What evidence did the researchers find to corroborate that thesis? Well, by their own admission: none. Instead, they inserted a bunch of hypothetical assumptions into an academic model, turned the crank a few times, pushed a couple buttons, and then, there it was: a theoretical conclusion based on the interplay of theoretical inputs.

Of course, as with all modeling studies, the quality of what you get out depends entirely on the quality of what you put in.  In this case, it appears the assumptions weren’t very good at all.  To wit:

Bad Input #1: Study significantly inflates expected development scenarios

Before discussing the individual inputs the study gets wrong in developing its risk assessment, its worth noting it significantly inflates natural gas development expected in New York state.  Specifically, it assumes that 40,000 wells will be developed in the Empire State.  From the study:

Given typical well spacing in the Marcellus Shale (10), if only 10% of the region is developed, this would equate to 40,000 wells. (page 10)

However, that isn’t at all what’s likely to happen according to experience in Pennsylvania and data from the New York Department of Environmental Conservation. According to the Pennsylvania Department of Environmental Protection approximately 5,000 wells have been developed in Pennsylvania since 2008. The SGEIS (executive summary, page 4) in New York has similar data noting, “The Department has determined, based on industry projections, that it may receive applications to drill approximately 1,700 – 2,500 horizontal and vertical wells for development of the Marcellus Shale by high-volume hydraulic fracturing during a peak development year. An average year may see 1,600 or more applications. ”  The SGEIS data is a liberal assessment and merely represents applications; the number of wells actually developed during a given year would likely be lower.

Bad Input #2: Study misses the mark on how water is being treated in the Marcellus Shale

The study declares that significant risks exist due to water treatment practices in use in the Marcellus Shale.  The problem? The study reaches this conclusion using an operating environment that simply doesn’t exist.  From the study:

Although some well operators recycle and reuse hydraulic fracturing fluids for multiple wells, most operators do not due to the cost of separation and filtration. Instead, the used hydraulic fracturing fluid is transported to wastewater treatment facility and discharged to streams. (page 6)

View of the Finger Lakes in New York

That kind of assertion would be news to Michael Krancer, PA DEP Secretary, who specifically outlawed that practice in May 2011, which, as reported by the Harrisburg Patriot News, resulted in a “sea change” of operations in Pennsylvania.  In fact, according to reports, most of the major operators in the Marcellus Shale are now recycling upwards of 90 percent of the water used in their operations.  This is a significant increase from the industry-wide 70 percent recycling rate achieved in 2011.  Of course, both of these rates far exceed the 16 percent assumed by Stony Brook researchers used in reaching their conclusion.

That’s an interesting omission, especially considering the study’s researchers themselves said this about reuse and recycling:

if the fluid is reused it was assumed there are no disposal losses for the well and any further losses are accrued by the next well. (page 4)

Of course, if the actual facts about water recycling in the Marcellus were used for this study, its conclusions would’ve been far different; but perhaps the study’s authors knew that.

Bad Input #3: Bad assumptions and data on natural gas well failures

In addition to ignoring Class II underground injection wells as a disposal option, the authors curiously rely on studies of those wells to determine casing failures of producing natural gas wells. The authors also rely on a limited study of 43 Class I underground injection wells, and cases of methane migration to determine their assumed well casing failure rate.  From the study:

 Regarding the likelihood of a well casing failure, a commonly cited statistic(10) originated with an American Petroleum Institute (API) report(37) that estimates the absolute risk of contaminating an underground source of drinking water from a Class II (oil & gas) injection well as between 2 × 10−5(1 in 50,000) and 2 × 10−8(1 in 50 million) well-years. The report uses historical well failure rate data and is based on the simultaneous failure of multiple well casings and fluids moving between the deep injection reservoir and surface aquifer.(38) It has been argued(10) that the API study serves as an upper bounds for well failure risk because wastewater injection wells continuously operate at higher than the geologic formation pressure whereas hydraulically fractured wells only operate above the formation pressure for a few days during construction… Although the sample size was only 43 wells, a subsequent report by the Underground Injection Practices Council (UIPC)(40) finds a 2% leak rate into underground sources of drinking water for Class I wastewater injection wells. The Pennsylvania Department of Environmental Protection (PA-DEP) found 52 separate cases of methane migration in a five-year period ending in 2009.(41) (Page 5)

Of course, anyone with even a passing knowledge of oil and gas operations knows that the mechanics and operations of  Class I and Class II underground injection wells are entirely different from working oil and natural gas wells. This is true both in the viscosity of what is being injected, as well as the pressure of injection and the duration of that pressure, among countless other variables.  An even more glaring error occurs when the authors reference 52 methane migration cases in Pennsylvania, then use this number  to “approximate” a well failure rate in the Commonwealth.  The only problem? Pennsylvania operates under presumed liability, meaning that if a baseline water sample wasn’t conducted prior to natural gas development, liability for methane migration falls on the operator.  With at least one study showing that pre-development tests found methane present in 78 percent of the water wells sampled. This makes it pretty clear that every one of the Stony Brooks study’s assumption in regards to well casing failure is inaccurate.

Had the authors been interested in using a well failure rate from experience they didn’t have to look too far; including an Aug. 2011 report issued by the Ground Water Protection Council (GWPC), a study that draws on field data and case descriptions from regulators in two of the most heavily drilled states in the country: Texas and Ohio. According to that study, more than 220,000 oil and natural gas wells were drilled and completed (fractured) in these two states over the past 25 years, 16,000 of them horizontal wells targeting deep shale formations. According to the report, more than 34,000 wells were drilled and completed in Ohio over a 25-year period from 1983 to 2007. In total, 184 incidents were recorded over that span in which oilfield activities – all categories – were found to have contributed to an adverse impact to groundwater. That’s one incident for every 184 wells drilled. Break the numbers down further, though, and you find that of those 184 incidents, only 12 were related to failures of, or gradual erosions to, casing or cement. That’s one recorded incident for every 2,833 wells drilled, representing a failure rate of 0.03 percent.

Another careless mistake the authors made was citing that “high salinity and dissolved solids in Appalachian rivers have been associated with the disposal of Marcellus Shale fracturing wastewater…”  This assertion was made based on an article that never made such a correlation.  In fact, written during preliminary investigations the article listed four possible sources of contamination.  Further review in the watershed revealed the true culprit was not Marcellus Shale development, but rather a combination of low stream flows and acid mine drainage impacting the watershed.

Of course, studies seeking to generate negative headlines often use flawed assumptions to reach skewed findings that paint a different picture than what is actually occurring.  This seems to be a key tenet of studies seeking to disparage the responsible development of natural gas from shale.  However, in this case, it appears the study was so recognizably flawed that it was incapable of achieving the task for which it was likely developed: generating negative headlines.