Variation in Methane Emission Rates from Well Pads in Four Oil and Gas Basins with Contrasting Production Volumes and Compositions

Transcription

1 SUPPLEMENTAL INFORMATION Variation in Methane Emission Rates from Well Pads in Four Oil and Gas Basins with Contrasting Production Volumes and Compositions Anna M. Robertson, + Rachel Edie, + Dustin Snare, Jeffrey Soltis, + Robert A. Field, + Matthew D. Burkhart, + Clay S. Bell, # Daniel Zimmerle, # Shane M. Murphy +, * Department of Atmospheric Science, University of Wyoming, 1000 E University Ave, Laramie, Wyoming 82071, USA # Energy Institute and Mechanical Engineering, Colorado State University, 430 N College, Fort Collins, CO 80524, USA All4 Inc., Kimberton, Pennsylvania 19442, United States * Corresponding Author: 1000 E University Ave, Laramie, WY, USA 82071; phone: ; Shane.Murphy@uwyo.edu S1

2 CONTENTS 1.0 OTM 33a Methodology S3 1.1 Mobile Lab Setup S3 1.2 Measurement Summary S3 1.3 Calculation of Emission Mass Flux.. S4 1.4 Data Flags. S5 1.5 Controlled Test Releases and Calculation of 95% Confidence Interval. S5 1.6 OTM 33a Data Provided by Brantley et al. S9 1.7 Advantages and Disadvantages of OTM 33a Methodology S Bootstrapped Natural Gas-Normalized Methane Average.. S TNMA Emissions vs. Water Fraction. S Measurement Summary for Each Basin. S Excluded Duplicate Measurements.. S FIGURES AND TABLES Figure S1: University of Wyoming Mobile Air Quality Lab S3 Figure S2: Illustration of our application of the OTM 33a method. S4 Figure S3: Gaussian fit of measured test release errors. S6 Figure S4: Throughput-normalized methane emissions for the DJ Basin, sorted by year.. S9 Figure S5: Bootstrapped TNMA emissions, binned by water fraction S12 Table S1: Summary of study dates and number of measurements performed. S3 Table S2: Summary of test release measurements S6 Table S3: Median Natural Gas-Normalized emissions for each basin. S11 Table S4: Summary of successful measurements made in each basin.. S13 Table S5: Summary of duplicate measurements that were excluded S19 S2

3 OTM 33a Methodology 1.1 Mobile Lab Setup B C A Figure S1 - University of Wyoming Atmospheric Science Mobile Air Quality Lab. Instruments indicated on the mast are: A) 3D sonic anemometer, B) 2D compact weather station, and C) inlet Measurement Summary Table S1 Summary of dates and number of measurements performed in each basin, including those from Brantley et al. 1 University of Wyoming Study Dates UGR DJ Uintah FV 2014 April & June 2014 November 2015 April - May 2015 September August October 51 Brantley et al. (2014) Study Dates Total Number of Measurements (Number from Brantley) 2010 July 2011 July - August (68 Brantley) S3

4 Calculation of Emission Mass Flux Inverting the mixing ratio measurements into a mass emission rate (Q) is calculated using the Gaussian plume dispersion equation: Q = 2π U σ y σ z C peak The symbol U represents the mean wind speed during the measurement. The estimated dispersion parameters ( y, z), are found in a look-up table using distance from the source and calculated atmospheric stability classes following the Pasquill-Gifford method, 2 see Figure S2. The atmospheric stability classes can range from 1 in very unstable conditions to 7 in very stable conditions, and depend on the horizontal standard deviation of the wind and the turbulent intensity. The normalized peak concentration (Cpeak) is derived from the Gaussian distribution fit to the measurements. The mass emission rate is assumed to be from a single collective point source (i.e., all emissions from a well pad mixed into a single plume) y z σ y σ z x Ū Figure S2 Illustration of our application of the OTM 33a method with the source at the intersection of the axes and the University of Wyoming mobile lab platform stationed meters downwind from the source. The symbol U is the mean wind speed during the measurement, y is the horizontal spreading parameter, and z is the vertical spreading parameter. 77 S4

5 Data Flags During analysis using the EPA methodology, measurements were categorized as a 1, 2, or 3 based on the number of data quality flags they raised. The quality of the data is based on several meteorological factors (e.g., wind speed, wind variance, and atmospheric stability), combined with physical parameters (e.g., distance from the source), which in turn determine how well the measured emissions can be represented as a distinct Gaussian plume. The EPA OTM 33a defines the categories as follows: Category Number of Flags Category 1 data are generally only excluded when there are outside factors noted during further analysis (e.g., multiple plumes measured during a transect of a possible source). Data flagged as a category 2 can have some plume shape and wind field errors, but if outside factors don t disqualify the data, they are considered valid measurements. Data in category 3 are considered invalid until further analysis can be performed and outside factors are considered. If low methane was the main reason for the data quality flags and the total number of flags raised was below 15, data was validated and included. If there was a significant number of flags raised in addition to low methane, the data was tagged as exceeding the data quality flag threshold and was excluded from further analysis Controlled Test Releases and Calculation of 95% Confidence Interval The analysis of our test release data set is summarized as follows: our mathematical standard deviation (one σ) is 24.4%, and a Gaussian fit of our data (see Figure S3) has a σ value of 28.2% (R = 0.98), centered at -9.97% (low measurement bias). Therefore, we conclude that our 95% confidence interval is +/- 56%, with a -10% bias. These values are based off of 109 test release data points that passed data quality flag thresholds, see Table S2, 90 of which were performed by the EPA. 1,3 The percent error was calculated by subtracting the known emission from the measured emission and dividing by the known emission. S5

6 Figure S3 Gaussian fit of measured test release errors. Our 95% confidence interval from the actual value measured is ±56% with a -10% bias. Overall, the University of Wyoming performed 23 test releases over 3 different days, with wind speeds varying from 2 9 m/s, distances of m, and controlled methane release rates of kg/h. Out of the 23 controlled releases performed, 19 passed the data quality threshold (based on the number of flags each measurement raised) described in SI section 2.4. The controlled releases were performed at the Christman Airfield in Fort Collins, CO, which is a flat, unobstructed field. The EPA performed a total of 107 controlled release measurements, 90 of which passed the data quality threshold and were included in the error analysis above. The EPA performed theirs at varying site locations over several different days, in relatively obstruction-free, open areas that more accurately simulated field measurement conditions (for oil and gas field measurements in western and Texas basins). Their controlled fluxes of methane ranged from kg/h, wind speeds from 1 7 m/s, and distances of m. Table S2 Summary of test release information performed by the University of Wyoming (first 19 measurements), and the EPA (last 90 measurements). Date (MDDYY) Distance (m) 3DS 2D Wind Speed (m/s) Approx Release Rate (kg/h) OTM Emission Estimate (kg/h) Percent Error S6

7 EPA Test Releases Date (#MMDDYY) Distance (m) 3DS 2D Wind Speed (m/s) Approx Release Rate (kg/h) OTM Emission Estimate (kg/h) Percent Error S7

8 S8

9 OTM 33a data provided by Brantley et al. To gain a more robust dataset for the DJ basin, we combined the data we collected with the OTM 33a data collected by Brantley et al., 1 adding 68 measurements. Measurements provided by Brantley et al. were collected in 2010 and 2011, compared to our data which was collected in However, analysis of year-to-year emissions showed that emissions did not change significantly over that time (Figure S4). The original data also contained several duplicate measurements at a single site. To prevent this from biasing any results, a random number generator was used to pick a single, representative measurement for that site Figure S4 - Box plots of throughput-normalized methane average emissions for the DJ Basin, sorted by the year measurements were made. The 2010 and 2011 data were provided by Brantley et al., and the 2014 measurements were made during the current study. On each box the central line represents the median, the edges of the box are the 25 th and 75 th percentiles, and the whiskers are 95% CI. Number of measurements made in each year are: 21 in 2010, 47 in 2011, 16 in S9

10 Advantages and Disadvantages of OTM 33a Methodology The OTM 33a methodology has several notable advantages and disadvantages over other current methane emission estimate methodologies, notably aircraft, tracer, and on-site measurements. Several aircraft studies have demonstrated that actual emissions exceed inventories for certain oil and gas fields, but are not able to address which operations or processes are responsible for excess emissions. Component-by-component, i.e. on-site, studies provide key information on individual equipment emissions but are costly, and require site access and operator cooperation possibly skewing samples. On-site is also unable to directly measure some sources on the well pad if safe access is not possible. Tracer release methods provide accurate measurements by using tracer to methane correlations, but measurements can be timely taking anywhere from half an hour to several hours, limiting sample size. OTM 33a, on the other hand, is able to identify the magnitude of individual sources throughout the field, can measure all emission sources from a single combined plume, does not require site access, and measurements take no more than a half an hour to complete. The method also requires only one vehicle and two operators. However, like any method, OTM 33a does have limitations and disadvantages. The inlet needs to be at roughly the same height as the source to be best aligned with the center of the plume, so hilly terrain and sparse access roads can make finding a downwind location to measure from difficult. Obstructions between the inlet and source, such as forested areas, can modify the wind field and bias the measurement. 4 Also, since measurements must be made relatively close to the source ( m) the primary disadvantage with this method is the inability to measure plumes with high vertical velocities (manual unloadings, stack emissions), since the majority of the plume will be above the inlet leading to large underestimations S10

11 Bootstrapped Natural Gas-Normalized Methane Average Natural gas-normalized methane average (NGNMA) is the amount of gross natural gas produced at the well pad that is emitted to the atmosphere, and is calculated as: NGNMA (%) = measured methane emission rate (mcfd) gross natural gas produced (mcfd) 100 Our measured methane emission rate (X) has been converted from kg/h to mcfd by: 174 X mcfd = X kg h 24 h day 1 mol CH ft3 1 mcf kg 1 mol CH ft Where the volume of one mole of an ideal gas (CH4 in this case) is found by using the same standard conditions used to calculate the monthly gas production values reported by each basin s respective oil and gas conservation commission (temperature = 60 o F, and pressure = psia). Table S3 lists the average NGNMA for each basin Table S3 Median percent of gross gas production emitted and 95% CI for each basin. Basin Median % of Gross Gas Produced Emitted (95% CI) Uintah Gas Wells 2.48 ( ) Uintah Oil Wells 25.8 ( ) DJ 1.59 ( ) UGR 0.17 ( ) FV (no MU) 0.08 ( ) S11

12 TNMA Emissions vs. Water Fraction Figure S5 - Box plots of bootstrapped TNMA emissions, binned by water fraction. On each box the central line represents the median, the edges of the box are the 25 th and 75 th percentiles, and the whiskers are the 95 th percentiles. Bins are , , , > 0.6. The number of measurements on which the bootstrap calculations were based for each bin are: UGR 13, 14, 16, 8; DJ 65, 17, 0, 0; Uintah 0, 0, 0, 11; FV 35, 8, 0, S12

13 Measurement Summary for Each Basin Table S4 Summary of successful measurements made in each basin. Columns correspond to basin ID, throughput-normalized emissions (%), OTM-33a emission rate estimate, gas production reported for that well pad during the month of measurement (MCFD), methane production for that well pad, water production reported for that well pad during the month of measurement, oil production reported for that well pad during the month of measurement, any notes for that well pad such as maintenance and whether the well pad was determined to be zero based on transect (0 BOT) during the FV campaign, and a reference ID for the sites measured in FV. OTM-33a Emission Estimate Gas MCFD Methane Water Oil Basin % Emitted Note FV BOT 1 FV BOT 2 FV BOT 3 FV BOT 4 FV BOT 5 FV BOT 6 FV BOT 7 FV BOT 8 FV BOT 9 FV BOT 10 FV BOT 11 FV BOT 12 Manual FV Unloading 13 FV FV FV FV FV FV FV Plunger FV Lift 21 FV FV FV FV FV FV Ref. ID S13

14 Basin % Emitted OTM-33a Emission Estimate Gas MCFD Methane Water Oil FV FV FV FV FV FV FV FV FV FV FV FV FV FV FV FV FV FV FV FV FV FV FV FV FV FV Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Note Ref. ID S14

15 Basin % Emitted OTM-33a Emission Estimate Gas MCFD Methane Water Oil Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah Uintah DJ N/A DJ N/A DJ N/A DJ N/A DJ N/A DJ N/A DJ N/A DJ N/A DJ N/A DJ N/A DJ N/A DJ N/A DJ N/A DJ N/A DJ N/A DJ N/A DJ - Brantley DJ - Brantley Note Ref. ID S15

16 Basin % Emitted OTM-33a Emission Estimate Gas MCFD Methane Water Oil DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley Note Ref. ID S16

17 Basin % Emitted OTM-33a Emission Estimate Gas MCFD Methane Water Oil DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley UGR UGR UGR UGR UGR UGR UGR UGR Note Ref. ID S17

18 Basin % Emitted OTM-33a Emission Estimate Gas MCFD Methane Water Oil UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR UGR S18 Note Ref. ID

19 198 Basin % Emitted OTM-33a Emission Estimate Gas MCFD Methane Water Oil UGR UGR UGR UGR UGR UGR UGR Note Ref. ID Excluded Duplicate Measurements Table S5 Summary of duplicate measurements that were excluded in each basin. For each well pad, we picked a single representative measurement, so as not to bias the measurement error or final statistics, by using a random number generator. Columns correspond to basin ID, throughput-normalized emissions (%), OTM-33a emission rate estimate, gas production reported for that well pad during the month of measurement (MCFD), methane production for that well pad, water production reported for that well pad during the month of measurement, oil production reported for that well pad during the month of measurement, and any notes for that well pad such as maintenance and whether the well pad was determined to be 0 BOT during the FV campaign. OTM-33a Emission Estimate Gas MCFD Basin % Emitted Note Ref. ID FV FV S19 Methane Water Oil Plunger FV Lift 47 FV FV DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley DJ - Brantley Plunger Lift 50

20 REFERENCES (1) Brantley, H. L.; Thoma, E. D.; Squier, W. C.; Guven, B. B.; Lyon, D. Assessment of methane emissions from oil and gas production pads using mobile measurements. Environ. Sci. Technol. 2014, 48(24), (2) Turner, D. B. Workbook of atmospheric dispersion estimates. Publication No. 999-AP-26, USEPA, Research Triangle Park, North Carolina, U.S.A, (3) U.S. EPA, Other Test Method (OTM) 33 and 33A Geospatial Measurement of Air Pollution-Remote Emissions Quantification Direct Assessment (GMAP-REQ-DA). 2014; (4) Rella, C.; Tsai, T. R.; Botkin, C. G.; Crosson, E. R.; Steele, D. Measuring emissions from oil and natural gas producing well pads in the Barnett Shale region using the novel mobile flux plane technique. Environ. Sci. Technol., 2015, 49, , DOI: /acs.est.5b S20