1. Improving Oil & Gas Emissions Tool Inputs Using Industry Surveys and Permit Data Mark Gibbs Environmental Programs Manager Emissions Inventory Section Air Quality Division, Oklahoma DEQ National Oil and Gas Emissions Committee Monthly Call and Industry Outreach November 12, 2015

2. Two Examples of How States Can Improve Tool Estimates 1.Oklahoma DEQ’s Revisions to Pneumatic Device Emissions for the 2011 Modeling Platform. 2.Spatial Heterogeneity in Gas Compositions from Oklahoma DEQ Permit Applications for Wellhead Facilities. 2

3. Source CategoryVOC (TPY) Artificial Lift87 Associated Gas1,295 Condensate Tanks20,944 Crude Oil Tanks25,825 Dehydrators5,829 Drill Rigs162 Fugitives32,633 Gas-Actuated Pumps15,932 Heaters558 Hydraulic Fracturing16 Lateral/Gathering Compressors484 Liquids Unloading20,658 Loading Emissions4,823 Mud Degassing11,094 Pneumatic Devices86,614 Produced Water0 Well Completions1,917 - Natural Gas well completions 334 Wellhead Compressor Engines978 Total230,183 Oklahoma 2011 NEI Oil & Gas Area Emissions From: National Oil & Gas Emissions Tool v 2 Pneumatic devices are apparently the largest source of VOCs. 3

4. Pneumatic Devices The single largest source category (for VOC emissions) for this sector. Not previously identified as a separate emissions unit in permits. Needed to better characterize this source. ODEQ proposed requiring permitted facilities to report all pneumatic devices for the 2014 point source inventory. OIPA members suggested a field study as an alternative to generate better data. 4

5. Initial Pneumatic Emissions Data (2011 O&G Tool) From the CenSARA 2012 study. Developed by ENVIRON based on surveys sent to companies responsible for 80% of the production in the basins investigated. Results augmented by professional judgment and previous work for other basins. 5

6. Criticism of the Initial Pneumatic Device Data Set Low response rate to operator surveys. – Anadarko Basin: 7 responses – Arkoma Basin: 6 responses Not based on field observations. Too much reliance on the 1996 GRI study for bleed rates. Did not distinguish between normal gas venting (for proper device operation) and malfunctions in the report and subsequent analysis. 6

7. OIPA Study Methodology Eight companies participated 21 Oklahoma counties represented Facilities from four of the seven CenSARA basins included Field study: 172 oil and gas wellhead facilities (205 wells) Old (first production prior to 2000) and new facilities included 70 oil wellhead facilities and 92 gas wellhead facilities 10 wellhead facilities had no pneumatic controllers on site 7

8. Low BleedIntermittentHigh Bleed Bleed Rate (scfh)3.2 4.0 (Anadarko) 8.6 (Other Basins) 13.1 (Arkoma) 32.1 OIPA Average Counts and Emissions by Site Type (680 natural gas pneumatic controllers: 659 intermittent vent controllers, 21 continuous bleed controllers) OIPA, November 2014, Pneumatic Controller Emissions from a Sample of 172 Production Facilities http://www.oipa.com/page_images/1418911081.pdf CenSARA Device Counts and Bleed Rates ElementAnadarkoArkomaOther Low Bleed devices/well1.110.990.94 High Bleed devices/well1.010.991.20 Intermittent devices/well0.951.200.86 8

9. Intermittent Vent Controllers CenSARA StudyOIPA Study Intermittent vent controllers represented 38% of all pneumatic controllers Intermittent vent controllers represented 97% of all pneumatic controllers 9 Intermittent vent controllers are far more prevalent than continuous bleed controllers. Expect this to be confirmed by the GHGRP Data. Emissions per well (from this source category) expected to be substantially lower.

10. OIPA Pneumatic Controller Emissions Histogram, 717 scfh gas in total OIPA, November 2014, Pneumatic Controller Emissions from a Sample of 172 Production Facilities http://www.oipa.com/page_images/1418911081.pdf 10 But What about the Fat Tails?

11. New Methodology (For Revisions to the 2011 Modeling Platform) Allocate emissions into two different SCC categories: 1)Emissions due to normal operations (pneumatic device SCCs) 2)Emissions due to malfunctions (fugitive “other” SCCs) 11

12. Properly Functioning Controllers Vent rate: 1.05 scfh (the arithmetic mean of all observed devices from the OIPA study) Devices per well: 3.6 (the arithmetic mean for all well types from the OIPA study) All other parameters (e.g., VOC weight percent, etc.) were kept identical to the original values from the CenSARA study 12 Apply to All Wells Statewide

13. Malfunctioning Controllers Vent rate: 2.0 scfh (Average additional rate for all wells assuming 3% are highly emitting at 50 scfh; this approach was informed by the results of the Allen et al. study.) Devices per well: 3.6 (the arithmetic mean for all well types from the OIPA study) All other parameters kept identical to the original values from the CenSARA study 13 Apply to All Wells Statewide

14. Net Changes in Pneumatic Controller Emissions 14 SCCDescription VOC Emissions (TPY) 2011 NEI v.2 Corrected Value Net Change 2310010300 Oil Production Pneumatic Devices 14,8596,513-8,346 2310021300 On-Shore Gas Production Pneumatic Devices 69,6046,390-63,214 2310023300 On-Shore CBM Production Pneumatic Devices 2,150158-1,992 2310021506 On-Shore Gas Production/ Fugitives: Other 10,89423,066+12,172 2310011506 On-Shore Oil Production/ Fugitives: Other 012,406+12,406 2310023516 On-Shore CBM Production/ Fugitives: Other 135436+301 Totals 97,64248,969-48,673

15. Conclusions to the Work on Pneumatic Controller Emissions First attempt to meaningfully estimate statewide emissions of malfunctioning devices This approach (while still conservative) yields significant reductions in emissions from pneumatic controllers. Industry cooperation and input is vital. Additional data (e.g., GHGRP device counts, new field studies, accounting for stripper wells) can be fed into the tool framework, further improving tool estimates. 15

16. Spatial Heterogeneity in Gas Compositions from Oklahoma DEQ Permit Applications for Wellhead Facilities 16

17. Methods Description Mined ODEQ permit applications for unique sales gas composition analyses Included counties that had at least 10 unique analyses Gas Composition: Weight % VOC ―Calculated ODEQ Weight % VOC averages for each county and compared those to Weight % VOC basin averages used by the O&G Tool VOC Emissions ―Ran the Oil & Gas Tool with a “No Change” scenario and an “ODEQ” scenario, which used spatially heterogeneous gas compositions ―Collected VOC emissions (TPY) output from the O&G Tool for both Pneumatics & Fugitives categories 17

18. 18 Weight % VOC Used For Pneumatic & Fugitive Emissions From EPA Oil & Gas Tool 23% 14% 3% 13% 14%

19. 19 Gas Analyses From Mined Permits For Wellhead Facilities

20. 20 Unique Gas Analyses Per County

21. 21 Weight % VOC Used For Pneumatic & Fugitive Emissions From Oklahoma DEQ Mined Permits

22. 22 Weight % VOC Used for Pneumatic and Fugitive Emissions From EPA Oil & Gas Tool and Oklahoma DEQ Mined Permits

23. Weight % VOC Used for Pneumatic and Fugitive Emissions From EPA Oil & Gas Tool and Oklahoma DEQ Mined Permits 23

24. VOC Emissions (TPY): State wide Oklahoma VOC emissions for Pneumatic Devices and Fugitives for the “No change” scenario and the “ODEQ” scenario of the Oil & Gas Tool Well TypeNo Change ScenarioODEQ Scenario Pneumatics Oil:14,85913,908 Gas:69,60471,831 CBM:2,1502,165 TOTAL:86,61487,904 Fugitives Oil:6,8586,727 Gas:25,77526,208 CBM:1,1661,017 TOTAL:33,79933,952 24

25. VOC Emissions (TPY): County Cases CountyNo ChangeODEQDifference (ODEQ – No Change) Beaver535398854532 Roger Mills446644682 Washita18911890 Woods33241861-1463 Pneumatics CountyNo ChangeODEQDifference (ODEQ – No Change) Beaver175140332282 Washita62365734 Stephens12211166-55 Pittsburg1443197-1246 Fugitives 25

26. 26 Total VOC Emissions (TPY) From Pneumatics and Fugitives Oklahoma DEQ Mined Permits EPA Oil and Gas Tool

27. 27 Difference in VOC Emissions (TPY) From Pneumatics and Fugitives Oklahoma DEQ Mined Permits – EPA Oil & Gas Tool

28. Overall Conclusions Industry cooperation and input can have a major impact on refining estimates. Detailed field studies can improve our understanding of key emission sources. Spatial heterogeneity can be significant at the county level. Need more data from multiple sources. 28

29. 29 Questions on These Projects?