Letter to the Editor

Letter to the Editor

Dear Dr. Rao,

Staff at the Texas Commission on Environmental Quality (TCEQ) read the above-referenced paper and found notable technical issues. While we have many questions about the study's modeling, inputs, and assumptions, the three technical issues below stand out.

Flare Emissions

The modeled flare, described as receiving natural gas in the abstract, was given a significant amount of propylene (a highly reactive volatile organic compound (VOC) ¹ that can increase ozone production efficiency). No evidence or citation was presented to substantiate the addition of propylene to the modeled flare's input stream at a natural gas facility. Based on TCEQ analyses, the majority of VOC emissions at oil and natural gas facilities are flash gas emissions, which occur when oil or condensed natural gas hydrocarbon liquids are reduced to atmospheric pressure after extraction (TCEQ, 2011a). Flash gas emissions have been sampled from storage tanks located at natural gas and oil wells at approximately 70 sites statewide. None of the lab analyses for these sites reported detectable quantities of highly reactive VOCs, such as propylene (Environ, 2010; TERC, 2009; ERG, 2009). Similarly, the EPA SPECIATE software's emissions profile for oil and gas production activities does not contain these compounds (EPA, 2012). The author's addition of propylene to an oil and gas service flare appears highly unrealistic, even for an emissions event. The author then fails to attribute the modeled ozone production from these flare emissions to his addition of propylene in the results and/or conclusions.

Modeling

The author modeled two self-described hypothetical emission cases for two hours and then extrapolated the results to potential impacts to eight-hour ozone concentrations. It is not appropriate to assume that two hours of plume modeling will adequately represent eight-hour ozone impacts from all oil and gas sources in the Barnett Shale. In addition, the paper offers no comparison to actual observations or other performance evaluation to determine the adequacy of this relatively unknown model.

Statements from the “Implications” and “Summary and Conclusion” sections suggest that this study's temporal and spatial resolution of emissions and models was significantly better than those used in State Implementation Plan (SIP) attainment demonstrations, though no comparison was presented. Although the model was run at a fine spatial resolution, this study did not improve the temporal or spatial resolution of emission inventories. In fact, the inventories were hypothetical and used emissions data that were significantly less chemically speciated than typical SIP modeling inventories. This study's hypothetical sources used continuous hourly emissions for the two-hour model run while SIP inventories vary hour-to-hour according to reported values where possible, or else according to source category-specific temporal surrogates.

TCEQ Regulations

The paper states, “Our findings suggest that improved regulation of the upstream oil and gas industry in nonattainment areas should include reporting of emissions events, and more aggressive deployment of control strategies such as vapor recovery to avoid flaring and the use of oxidation catalysts on stationary engines.” While this may be true for other states, the author failed to point out that the identified emissions reduction opportunities in the report have already been implemented by the TCEQ.

a.	In Texas all companies are required to record emissions events and are also required to submit reports to the TCEQ for emissions events above the reportable quantity thresholds detailed in 30 Texas Administrative Code 101.1 (TAC, 2011).
b.	The latest DFW State Implementation Plan revision requires vapor recovery for storage tank batteries that emit over 50 tons of VOC per year (TCEQ, 2011b).
c.	According to the 2009 TCEQ Barnett Shale Phase II emission inventory (TCEQ, 2011a), compressor engines with installed controls operated 87% of the horsepower-hours in the 9-country Dallas-Fort Worth (DFW) nonattainment area. As with DFW, compressor engines in the Houston-Galveston-Brazoria ozone nonattainment areas and eastern Texas have installed controls as required by 30 Texas Administrative Code Chapter 117 rules since being adopted in 2007 (TAC, 2007). Catalyst controls typically installed to meet the 30 Texas Administrative Code Chapter 117 rules can reduce acrolein and formaldehyde emissions by approximately 70%.

While running hypothetical modeling scenarios can be helpful in improving the understanding of the causes of and solutions to air pollution issues in Texas and elsewhere, researchers must exercise caution when extrapolating the results to the realm of public policy. Models to be used for ozone attainment demonstration modeling or point source dispersion modeling need to be approved by EPA before they can be used for regulatory purposes (EPA, 2007), such as assessing the ozone contributed by different sources. This requirement is maintained to ensure that models used to inform policy decisions have been widely vetted by the scientific and regulatory communities.

Thank you for this opportunity to comment.

Sincerely,

David Brymer, Director

Air Quality Division

Texas Commission on Environmental Quality

Notes

¹ Propylene; ethylene, 1,3-butadiene; and all isomers of butene are commonly defined as highly reactive VOCs.