by Vaclav Smil
If true, then shale gas would lose all advantage in terms of carbon intensity of its production, although higher combustion efficiencies in household furnaces, large boilers, and gas turbines would still give it an overall, but relatively slight, edge. In response to a critique of their findings, Howarth, Santoro, and Ingraffea (2012) stood by their conclusions, and a study of air measurements at the Boulder Atmospheric Observatory tower, done to assess emissions from Denver–Julesburg Basin, ended up with a similar emissions range: Pétron et al. (2012) concluded that about 4% (range of 2.3–7.7%) of raw methane is leaking from infrastructure associated with gas production, a rate at least twice as high as has been generally assumed by the industry. These studies received wide attention because they suggested that methane leaks during gas production may entirely offset climate benefits of natural gas (Tollefson, 2012). Alvarez et al. (2012) concluded that emissions higher than about 3.2% of gas extraction result in immediate net radiative forcing worse than coal when the two fuels are used to generate electricity.
But the claims of gas being a worse choice than coal were questioned as responses to Pétron et al. (2012) pointed out that the conclusion depends on the way the measured concentrations were attributed and interpreted (Cathles, 2012; Levi, 2013). Most notably, different assumptions about air mixing with the stock tank and gas well leakages would produce CH4 loss rates of just 1.5–2%. O’Sullivan and Paltsev (2012) reviewed emission data from about 4,000 horizontal fracked wells brought into production in 2010, and they highlighted the difference between potential and actual methane losses. While potential fugitive emissions averaged 228,000 kg CH4 per well, the use of flaring and reduced emissions during well completion (capturing the flowback gas) lowered the actual emissions to about 50,000 kg CH4 per well, much below some widely quoted estimates. This led them to conclude that hydraulic fracturing of shales has not substantially altered the overall intensity of greenhouse gas emissions from the US natural gas extraction.
Other studies confirmed the benefits of natural gas. Deutsche Bank Climate Change Advisors study of life-cycle greenhouse gas emissions used the USEPA’s adjusted methodology, and it found that, on the average, US natural gas-fired electricity generation emitted 47% less greenhouse gases than coal (Deutsche Bank Group, 2011). And a comprehensive reevaluation of life-cycle greenhouse gas emissions of shale gas, conventional gas, coal, and petroleum—based on the USEPA’s latest emission estimates and comparing the impact per MJ of fuel burned, per kWh of electricity produced, and per km driven for transportation services—concluded that that shale gas emissions are 6% lower than those from the conventional gas (but the difference is too small to make claim indisputable), 23% lower than from gasoline, and 33% lower from coal (Burnham et al., 2011a).
Another study published in the same year introduced three interesting comparisons of natural gas-switching scenarios (Alvarez et al., 2012). The study did not focus specifically on shale gas as it compared the benefits of switching from gasoline or diesel and from coal-fired to gas-fired electricity generation, but its conclusions provide fairly emissions threshold for the GWP benefit of such substitutions. If CNG were to reduce immediately climate impacts from heavy-duty vehicles, well-to-wheels CH4 leakage must be kept below 1% of total extraction, and new gas-fired plants would be better than efficient, new coal plants only if the leakage in the entire natural gas system (from well to delivery) is kept to less than 3.2% of total production. The study also called for a much better understanding of methane leakage from natural gas infrastructure, an appeal strengthened by a review of two decades of publications on natural gas emissions in the United States and Canada (Brandt et al., 2014).
On one hand, this metastudy concluded that actually measured emissions are consistently higher than those indicated by standard emission inventories and factors (with measured/inventory ratios commonly around two) and that a small number of what they call “superemitters” could be responsible for very large shares of the leaked gas. On the other hand, the authors concluded that high leakage rates suggested in some recent studies are unlikely to be representative of the entire natural gas industry (if so, the emission associated with natural gas production would exceed the observed total excess CH4 from all sources) and that assessments done for 100-year impact do not indicate any system-wide leakages large enough to negate climate benefits of substituting coal by natural gas.
This controversy will continue for years to come as even near-perfect measurements of emissions from a significant number of wells cannot be extrapolated with a high degree of confidence to quantify nationwide leakage or assumed to represent a similar group of wells to be completed in the near future. The latest inventory of US greenhouse gas emissions and sinks, published in a preliminary version in February 2014, puts the total 2012 emissions from the country’s natural gas systems at 6.2 Mt CH4 or 130 Mt of CO2e, with transmission and storage releasing about a third of the total and field production about 30% (USEPA, 2014). Emissions of about 6 Mt CH4 are equal to 8.5 Gm3 or to almost exactly 1% of gross natural gas withdrawals of 836.3 Gm3 in 2012. But just 2 months later, publication of a field study from the Marcellus shale region of Pennsylvania indicated some exceptionally high CH4 fluxes from natural gas wells (Caulton et al., 2014).
The study relied on an instrumented aircraft to measure emission rates in southwestern Pennsylvania in June 2012. While a regional flux of 2.0–14 g CH4/s/km2 was in the same range as bottom-up inventory (2.3–4.6 g CH4/s/km2), emissions from seven well pads in the drilling phase averaged 34 g CH4/s per well, or two to three orders of magnitude higher than estimated by the USEPA for that stage of gas production. The studied wells accounted for only about 1% of all wells in the region, but their emissions produced 4–30% of the observed regional flux, confirming the previous conclusion about “superemmiters.” These findings are in contrast with the latest US inventory of greenhouse gas emissions that shows CH4 releases from natural gas systems declining by about 4% between 2005 and 2012 even as the gas extraction rose by 33% (USEPA, 2014).
Brandt et al. (2014) concluded that the US methane emissions are 25–75% (best estimate about 50%) higher than the USEPA’s total but that higher emissions from hydraulic fracturing used in shale gas extraction account for only about 7% of the additional CH4. Moreover, system perspectives may show that downstream leaks are no less important than the field emissions. Studies of methane emissions from pipeline and other leaks in Boston indicate higher than expected rates, and reducing those losses may be a key component of making shale gas less of a factor in any climate change concerns (Kintisch, 2014).
Recent quantifications of CH4 releases from natural gas systems range from 0.6 to 11.7% of gas production for upstream (well sites) and midstream (processing) activities and from less than 0.1 to 10% for downstream (transmission, storage, distribution). But even if we assume that a high average of 5% were to apply to all natural gas production systems (and not just to hydraulic fracturing), then the 2013 global gas production of 3.37 Tm3 would release nearly 170 Gm3 (110 Mt) CH4. That would be nearly 60% higher than the previously cited best estimate of CH4 emissions attributable to natural gas—but the difference (40 Mt CH4) remains considerably smaller than the lasting range of uncertainty regarding the natural emissions of the gas from wetlands (140 Mt) and there are numerous opportunities to reduce leakage.
More studies are unlikely to narrow the existing uncertainty of emission rates, and disputes about the life-cycle warming potential of gas systems in general and shale gas extraction in particular will continue. What is not at all in dispute are wasteful high emissions of CO2 generated by gas flaring that have accompanied shale oil production in the United States. Development of the Bakken formation of North Dakota, the state whose crude oil output had increased more than 100-fold between 2003 and 2013 and that became the country’s second highest oil producer after Texas in spring 2012 (USEIA, 2014l), is the most visible proof of this waste. Extraction of Bakken shale oil is as
sociated with large volumes of liquid-rich natural gas, and construction of new gathering line and trunk pipelines has not kept pace with the rapid rate of drilling new wells and producing new crude.
As a result, nighttime satellite images show that gas flaring in northwestern North Dakota has created a patch of light whose intensity is not quite as bright as those of the nearest large cities, Minneapolis in Minnesota and Denver in Colorado—but whose size is considerably larger (Figure 7.7). But this wasteful and environmentally harmful reality is no generic and lasting indictment of hydraulic fracturing, merely a result of specific circumstances resulting in excessive temporary flaring that will be reduced (North Dakota’s new standards approved in July 2014 will require capture of 90% of all gas released during drilling) and eventually eliminated as new pipelines enter into service.
Figure 7.7 Flaring in Bakken.
© NASA.
Comparisons of natural gas losses associated with conventional drilling and hydraulic fracturing show no or minimal difference for most routine activities, provided that all of them are conducted with care, a condition that is not always present in early stages of resource extraction of which the rapid development of Bakken shale has been a notable example. And compared to often long-lived conventional wells, the exponential decline in well productivity in fractured shale formations will require more frequent drilling to maintain a given level of output and hence present more opportunities for methane leaks. But these challenges have technical solutions, and there are no insurmountable problems that would prevent methane emission from shale gas production to be comparable, or lower, than those from conventional operations.
And before leaving the topic of atmospheric emissions associated with natural gas production, I must also mention that in some area there may be also seasonal impacts associated with the releases of nitrogen oxides (NO and NO2) and volatile organic compounds. These air pollutants are the essential precursors of photochemical ozone formation, and Edwards et al. (2014) found that in Utah’s Uinta Basin this leads to winter ozone mixing ratios well in excess of present air quality standards.
7.3.3 Water Use and Contamination
Making clear-cut conclusions concerning the consumptive use of water, contamination of drinking water aquifers and injection of salt-laden water into deep wells is even more difficult than judging the atmospheric impacts. HVHF is much more demanding and has a much greater impact than the extraction of conventional natural gas resources, but water requirements by HVHF need careful qualifications. A simple estimate shows that water needed for HVHF adds up to a tiny share of total US water withdrawals. In the United States, usual ranges of overall water requirements are given either as 2–8 million gallons or, more narrowly, as 4–6 million gallons per well, that is, as little as 7.5 and as much as 33 Ml (or 7,500–33,000 m3), compared to 1 million gallons (3.8 Ml) for a vertical well. Obviously, fracking longer lateral wells increases water demand, generally by about 18,000 l for every additional meter (Fractracker Alliance, 2014).
Average of 15 Ml (15,000 m3) would be perhaps the most representative requirement if a single value is needed to make proper order-of-magnitude estimates of aggregate water use. For about 30,000 wells (maximum drilled recently), that would amount to about 450 Mm3 or less 0.1% of all US water use in 2005 (Kenny et al., 2009). The shares are similarly low even in regions of concentrated HVHF: in the Marcellus formation in Pennsylvania, water consumption for fracking is only about 0.2% of total annual water withdrawals (Vidic et al., 2013). The share may be lower in rainy watersheds, but it could be much higher in arid regions. In Texas, HVHF uses less than 1% of the state’s water withdrawals but with a large share of the state being so arid that may still tax local water supplies.
Yet another perspective is to consider water intensity of the most common alternative. Because coal will be the fossil fuel most commonly displaced by natural gas, it is instructive to compare water intensity of coal-based electricity generation with water intensity of shale gas-based generation. Scanlon, Duncan, and Reedy (2013) did precisely that for Texas (the state where concerns about water supply are particularly acute), and their conclusion is impressive: water saved by using gas-fueled CCGT instead of coal steam turbine plants is 25–50 times the volume of water used in hydraulic fracturing to extract the gas.
What makes the water use in fracking so demanding is that a requisite volume must be delivered to a single site within a brief period when fracking operations take place, that in many arid environments even a much smaller volume could not be secured from nearby sources, and that the delivery is nearly always an environmentally stressful experience. There are, of course, many single-well operations, but multiwell pads are becoming more common, and hence, the actual volume required at a single site may be commonly three to ten times as large (extreme range being 2-20 wells). Assuming the average requirement of 15,000 m3/well (mostly taken from streams and ponds, less frequently from municipal supplies and from recycled flowback liquid), a five-well pad site would need 75,000 m3. With 25 m3 of volume per truck, that would require 3,000 truck trips to a drill site, with noisy heavy diesel-powered trucks destroying rural roads (Figure 7.8).
Figure 7.8 Heavy truck carrying fracking liquid.
© Corbis.
Even more importantly, many water uses in modern society are nonconsumptive, and that includes the largest industrial use for the cooling of condensed water in thermal electricity-generating plants: except for a small share lost to evaporation, water is returned, slightly heated, to lakes, ponds, or streams. And while other uses are consumptive—evapotranspiration and leaching removes irrigation water from crop fields—that water remains part of the rapid global water cycle as it gets precipitated (close or far downwind) or reenters aquifers. In contrast, water used for fracking is not only consumed in the process but ceases to be a part of rapid water cycle. Most of the loss is immediate as usually at least 70% and commonly up to 90% of water used for fracking remains deep underground within shale formations. In many cases, the fate of this water is uncertain: it may be absorbed by shales, but in some settings, it may be also a potential contaminant of aquifers.
Once HVHF is completed and pressure is lowered, a portion of fracking liquid mixed with formation water—anywhere between 10 and 50% of the volume initially pumped into the well, with modal flows close to 10%—returns to the surface (most of this flowback takes place in the first 2–3 weeks) and has to be hauled away for recycling (necessitating hundreds of additional truck trips from a drill site). But, unlike water polluted with organic substances, flowback fracking liquid cannot be recycled infinitely (its concentration of salts becomes eventually too high, and sometimes, even its initial contamination is too severe to be handled by standard water treatment plants), and it has to be disposed of, usually by injections into deep wells (far below drinking water aquifers) and hence again taken out of water cycle. In addition, drilling of a single horizontal well can yield on the order of 100 t of organic-rich cuttings whose dumping would acidify affected waters and leach heavy metals; moreover, the cuttings may also have impermissibly high levels of radiation and must be disposed in secured landfills (USGS [United States Geological Survey], 2013).
Water dominates the volume of fracking liquid (typically 90%). If varieties of sand, the most common proppant (an ingredient introduced to keep fractures open and hence to enable gas and oil flow, and usually amounting to about 9% of fracking fluid mass) were the only other ingredient, then concerns about fracking and management of waste water would be much less challenging. But all fracking liquids contain numerous additives: about 750 substances have been used by the industry, and many fluids contain scores of different chemicals. Composition of fluids used in different regions is available from Halliburton (2014). Although their relative volume is small (typically <0.5%), their total volume used per fracking operation may be up to 150,000 l (150 m3).
Common additives include acids (helping to dissolve minerals and initiate ti
ny cracks), corrosion and scale inhibitors (methanol, formic acid), friction reducers (mostly petroleum distillates), gelling agents (to thicken fluid and help to suspend proppants, commonly guar gum and petroleum distillates), surfactants (to reduce the fluid surface tension, usually ethanol or methanol), pH adjusters, and biocides (FracFocus, 2014). Biocides (glutaraldehyde, ammonium chloride) are needed because stream or pond water contains bacteria whose growth could be enhanced by organics added to fracking fluid and result in emulsion problems, plugging, and souring (Biocides Panel, 2013).
Claims of contaminated drinking waters have come from a number of states (Arkansas, Colorado, New York, Pennsylvania, Texas, Virginia, Wyoming) and have been a major rallying point for the opponents of fracking. Reading their pamphlets and web pages, one would think that all methane is coming out of all faucets in the regions with HVHF. At the same time, it appears that some homeowners living less than 1 km from a shale gas well have a higher probability of having their drinking water contaminated with stray gases: that was the conclusion made by Jackson et al. (2013) after they analyzed 141 drinking water wells in northeastern Pennsylvania by correlating natural gas concentrations and isotopic signatures with proximity to shale gas wells. They detected methane in 82% of drinking water samples, and its average concentrations were six times higher in homes less than 1 km from natural gas wells, but, in a majority of cases, below the level that would require action for hazard mitigation.
