Characterizing emissions from natural gas drilling and well completion operations in Garfield County, CO

Abstract

This study was designed to characterize and quantify emission rates and dispersion of air toxics, ozone precursors, and greenhouse gases from unconventional natural gas well development activities in Garfield County, CO, located on top of a geological formation known as the Piceance Basin. Particular focus was placed on quantifying emissions of individual volatile organic compounds (VOCs) and methane during well drilling, hydraulic fracturing ("fracking"), and flowback. While some prior studies have measured VOC or methane concentrations near well development operations, ambient concentrations are strongly dependent not only on emission rates but also on sampling location and meteorological conditions, which greatly affect downwind dispersion and dilution. By characterizing emission rates directly, results from this study can be used to predict downwind concentration fields for any location of interest under a wide range of weather conditions. Emission rates were determined using a tracer ratio method (TRM). In this method, the rate of emission of a compound of interest (e.g., g s-1 of benzene) is determined as the product of a known tracer emission rate multiplied by the ratio of the background-corrected concentrations of the compound of interest and the tracer. Acetylene was selected as a tracer gas and its controlled release co-located with the main source of emissions on studied well pads. Real-time methane and acetylene concentrations and three minute integrated whole air sample canisters for VOC analysis were collected downwind of the release location. Meteorological data were collected at two heights (3 m and 10 m) near the well pad. Upwind acetylene, methane, and VOC concentrations were determined for background correction. The canisters were analyzed for a large suite of VOCs using gas chromatography with flame ionization detection. The study results provide novel information concerning emissions from natural gas drilling and completion activities in Garfield County, CO and are some of the first measurements of this type in any U.S. basin. Overall, 21 emission experiments were conducted from 2013-2015. Several sets of 2 to 5 canisters were collected at different times during each experiment, in addition to an upwind background sample per experiment. Using the TRM, each canister in the plume provides an independent measure of VOC emission rates. 28-48 VOCs are reported for each canister, along with real-time methane and acetylene data collected during each experiment. Using the TRM the emission rates of methane and individual VOCs are calculated and reported.The emission rates and field observations were used to conduct air dispersion (using the EPA's AERMOD model) simulations to: (1) evaluate AERMOD's accuracy in predicting observed, near-field dispersion of VOCs in Garfield County, CO and (2) predict concentration fields, as a function of emission rate, for dispersion of a hypothetical compound under a range of local meteorological conditions at a site with terrain similar to that observed in Garfield County. While not perfectly designed for prediction of the short-term concentration fields measured in the study, AERMOD did a reasonable job predicting the observed extent of dispersion across several field experiments. Moreover, emission rate ranges determined by activity type in this study can be used in a wide range of future simulations with AERMOD or other models to simulate downwind concentration fields relevant to understanding potential local health and air quality impacts associated with well development activities in Garfield County.