Page 2 of 5
Based on the study’s text (supplemental information was not available at the time of this pressrelease), we identified a number of methodological issues - including some of the samecriticisms launched against the EPA/GRI study - that render this study incomplete and problematic in terms of it being representative of well site methane emissions in the real world:1.
The study sites selected are not likely representative of typical gas development
A very small sample size
The study measured emissions from just 489 gas wells and only 27 hydraulic fracturingevents. These measurements represent just 0.11% of the total (conventional andunconventional) gas wells in the United States. EPA/GRI (1996) sampled a similarly smallsubset of the nation’s wells (0.14%).
Non-random choice of sites
The study’s results, much like those of the earlier EPA/GRI (1996) study, are based onevaluations of sites and times selected by the oil and gas industry rather than random andindependent sampling of sites. Thus, this study must be viewed as a best-case scenario, basedupon wells selected by industry, a party undoubtedly interested in a particular outcome (i.e.low methane loss from gas development).As stated in the paper,
“The uncertainty estimate does not include factors such asuncertainty in national counts of wells or equipment and the issue of whether thecompanies that provided sampling sites are representative of the national population.”
Unfortunately, these are exactly the most important parameters on which to base a trulyrepresentative, nation-wide assessment. And it is exactly the many co-authors of the paper who are employed by the gas industry who should have been in best position to know wellcounts, the equipment deployed at each, and whether the very few wells sites they used inthis paper were truly representative.
Type of gas wells sampled for flowback measurements is not clear and the results might say little about shale gas
Geological formation type has a significant impact on methane emissions during post-fracturing flowback. For example, industry data indicate
completion emissions from tight-sand and shale wells may be up to 10 times higher than those of coalbed methane orconventional wells
(Howarth, Santoro et al. 2011;
Pétron, Frost et al. 2012
; Karion, Sweeneyet al. 2013). However, Allen and colleagues do not address this critical point and it is unclear whether the 27 wells sampled during flowback are shale, tight-sand, coalbed methane, or acombination of well types. Based on the wide range of non-captured or flared emissionsreported, it does seem that both lower pressure wells (e.g. coalbed methane) and high pressure wells (e.g. shale) were included in the sampling, yet transparency of the distributionof the sampling across formation types is also omitted. If multiple formation types wereindeed sampled, simply averaging all measurements, without weighting them for differencesin formation type, biases the results.