National Oil and Gas Emissions Committee Monthly Committee

National Oil and Gas Emissions Committee Monthly Committee Call OKDEQ Procedures for Preparing Area Oil & Gas Emissions Estimates for the 2014 NEI May 2016 Mark Gibbs Environmental Programs Manager Emission Inventory Section Air Quality Division, Oklahoma DEQ

Oklahoma Oil & Gas NEI Submission History 2008 NEI 2011 NEI 2014 NEI Wellhead CENRAP Study Oil & Gas Tool + Aggregated Permitted Wells Midstream Aggregated Midstream Emissions Point Source Midstream Emissions Major Point Sources Type A & B Actuals Type A & B Potential 2

3

• Oil and Gas Tool + Aggregated Permitted Wells • Point Source Midstream Emissions • Type A & B Potential Point Sources 4

Oklahoma Area Source Oil & Gas Production Emissions 2014 O&G Production Tool Oklahoma Changes Point Source Inventories from Permitted Wellheads 2014 NEI Area Oil and Gas Submission to EIS 5

Oklahoma Modifications: What is Different? • OKDEQ Gas Composition: We used gas composition data for the 15 counties where we had sufficiently representative data • Pneumatic Devices: We incorporated the results of an OIPA study on pneumatic devices performed in Oklahoma (normal operations + malfunctions) • Wellhead Point Source Inventory Data: We incorporated point source inventory data from over 5, 000 oil and gas wellheads into our area source submission – This number of wells and their associated activity was removed from the Tool 6

Area Oil & Gas Emissions Estimation Tool • Goal: To capture a best estimate of criteria pollutant and HAP emissions from the Production oil and gas sector across the state of Oklahoma Involved many runs of the tool, and manipulation of the staging tables Main objective: prioritize getting all of the point source emissions into the final staging table over perfecting the point to nonpoint crosswalk (TONS matter more than SCCs) 7

Overview of this Presentation • Tool Scenario Runs: running the O&G tool with different source categories, basin factors, speciation factors, etc. • Procedures: adding emissions from the different scenario runs, and from the OKDEQ point emission inventory to the final staging tables • Final Comparison: comparing VOC & NOx for different versions of the tool • Key Observations: Differences between new, permitted wells and other wells 8

Procedure Snapshot (1) • HAP emissions: the Baseline Tool (without the activity subtractions for point sources) + OK gas composition data + OK pneumatic device bleed rates and malfunctions • We did not use our point source inventory to generate any HAP emissions 9

Procedure Snapshot (2) • Pneumatic device emissions: the Baseline Tool (without the activity subtractions for point sources) + OK gas composition data • Two runs: 1. Normal operation (avg bleed rate: 1. 05 scfh) 2. Malfunctions (fat tails → avg 2. 00 scfh additional emissions) • We did not use our point source inventory to generate emissions from pneumatic devices, because following the OIPA study we did not require these devices to be reported in the point inventory 10

Procedure Snapshot (3) • O&G Tool (for source categories other than pneumatic devices): Criteria pollutant emissions generated (with the activity subtractions for permitted wellheads) + OK gas composition data. • Point sources: Used data from our point source emissions inventory to generate criteria pollutant emissions from other sources (not pneumatic devices at permitted wellheads). – Facility-specific (or at least locally representative) gas composition and production data reflected in the inventory. – Required a complex crosswalk between point and nonpoint SCCs. 11

Comparison: OKDEQ Point Source Activity Data OKDEQ Permitted Wells TOTAL (HPDI) % Point Total Liquids (BBL) 55, 901, 885 117, 319, 665 48% Total Gas (MCF) 713, 638, 549 2, 162, 187, 694 33% # Oil Wells 3, 397 21, 060 16% # Gas Wells 843 31. 395 2. 7% # CBM Wells 66 3, 113 2. 1% • Note the range in coverage • OKDEQ permitted wells capture: almost 50% of the total liquids production, but only 16% of oil wells 33% of the total gas production and less than 3% of gas wells 12

Comparison: Activity Per Well Oil Wells Gas Wells CBM Wells Crude Oil (BBL/well) Associated Gas (MCF/well) Condensate (BBL/well) Natural Gas (MCF/well) OKDEQ Permitted Wells 16, 289 154, 286 674 223, 805 0 13, 042 Residual Wells 3, 400 17, 523 37 36, 389 15 8, 652 • Wells from the OKDEQ point source inventory are (in general): Newer wells Higher production wells 13

Tool Scenario Runs

Tool Scenario: Dec 2015, NEI run • “Dec 2015”: Version of the Tool released on December 4, 2015 (HPDI 2015 activity data & Subpart W data) • “NEI run”: Includes all area SCCs, OKDEQ local gas composition data (speciation factors for fugitives, gas -actuated pumps, liquids unloading, and pneumatic devices), and pneumatic device basin factor data for normally functioning pneumatic devices (OIPA, 2014) Dec 2015, NEI run (All Area SCCs) 15

Tool Scenario: Dec 2015, NEI run BASIN FACTOR - DATA_CATEGORY Number of No Bleed devices (COUNT) Bleed rate, No Bleed devices (SCF/HR/DEVICE) Number of Low Bleed devices (COUNT) Bleed rate, Low Bleed devices (SCF/HR/DEVICE) Number of High Bleed devices (COUNT) Bleed rate, High Bleed devices (SCF/HR/DEVICE) Number of Intermittent Bleed devices (COUNT) Bleed rate, Intermittent Bleed devices (SCF/HR/DEVICE) CURRENT_VALUE CURRENT_REFERENCE 0 0 0 3. 6 1. 05 OIPA_STUDY_2014 OIPA_STUDY_2014 OIPA Basin Factors: for normally functioning pneumatic devices. OKDEQ Speciation Factors: calculated only for counties with at least 10 unique sales gas composition analyses 16

Tool Scenario: Dec 2015, Malfunctions run • “Dec 2015”: Version of the Tool released on December 4, 2015 (HPDI 2015 activity data & Subpart W data) • “Malfunctions run”: includes ONLY pneumatic device SCCs, and basin factor data for malfunctioning pneumatic devices (OIPA, 2014) Dec 2015, Malfunctions run (For Pneu. Dev. ONLY) 17

Tool Scenario: Dec 2015, Malfunctions run BASIN FACTOR - DATA_CATEGORY Number of No Bleed devices (COUNT) Bleed rate, No Bleed devices (SCF/HR/DEVICE) Number of Low Bleed devices (COUNT) Bleed rate, Low Bleed devices (SCF/HR/DEVICE) Number of High Bleed devices (COUNT) Bleed rate, High Bleed devices (SCF/HR/DEVICE) Number of Intermittent Bleed devices (COUNT) Bleed rate, Intermittent Bleed devices (SCF/HR/DEVICE) CURRENT_VALUE CURRENT_REFERENCE 0 0 0 3. 6 2. 0 OIPA_STUDY_2014 OIPA_STUDY_2014 OIPA Basin Factors: for malfunctioning pneumatic devices. (Emissions from normally operating pneumatic devices were generated from the previous run. ) 18

Tool Scenario: Jan 2016, OKDEQ run • “Jan 2016”: Version of the Tool that released on December 4, 2015 (HPDI 2015 activity data, Subpart W data, plus OKDEQ point well data for point source subtraction). • “OKDEQ run”: includes all SCCs EXCEPT pneumatic devices, and OKDEQ local gas composition data (speciation factors for fugitives, gas-actuated pumps, and liquids unloading) Jan 2016, OKDEQ run (All Area SCCs EXCEPT Pneu. Dev. ) (Point Source Subtraction occurs) 19

Procedures

Procedure Flowchart • Running each Tool scenario generated three separate staging tables: Dec 2015, NEI run (All Area SCCs) Dec 2015, Malfunctions run (For Pneu. Dev. ONLY) Staging Table Jan 2016, OKDEQ run (All Area SCCs EXCEPT Pneu. Dev. ) (Point Source Subtraction occurs) Staging Table • The following procedures were implemented to manipulate/combine the output in these staging tables to form a final EIS submission 21

Procedure Flowchart: Step 1 Dec 2015, NEI run Dec 2015, Malfunctions run (All Area SCCs) (For Pneu. Dev. ONLY) Staging Table Jan 2016, OKDEQ run (All Area SCCs EXCEPT Pneu. Dev. ) (Point Source Subtraction occurs) Staging Table Delete all records for criteria emissions (except VOC from pneu. dev. ) Aggregate HAP Fugitive component SCC ems, then Append into new Fugitive - All Process SCCs Delete all ems records for Fugitive component SCCs STEP 1: Accounts for HAPS statewide, and VOC emissions from normally-functioning pneumatic devices Final Staging Table 22

Procedures: Step 1 • Delete all criteria pollutant records from the Emissions table EXCEPT for VOC emissions from pneumatic devices • Aggregate fugitive component emissions by well type, then append into new Fugitives: All Process SCCs Gas Wells CBM Wells Oil Wells Connectors 2310021501 2310023511 2310011501 Flanges 2310021502 2310023512 2310011502 Open Ended Lines 2310021503 2310023513 2310011503 Valves 2310021505 2310023515 2310011505 Other 2310021506 2310023516 N/A ↓ ↓ All Process 2310021509 2310023509 2310011500 • Delete all fugitive component SCC records in the Emissions and Emissions. Process table 23

Procedure Flowchart: Step 2 Dec 2015, NEI run Dec 2015, Malfunctions run (All Area SCCs) (For Pneu. Dev. ONLY) Staging Table Delete all records for criteria emissions (except VOC from pneu. dev. ) Update malfun. pneu. dev. BTEX ems to new Fugitive - All Process SCCs Aggregate HAP Fugitive component SCC ems, then Append into new Fugitive - All Process SCCs Append malfun. pneu. dev. VOC ems to new Fugitive - All Process SCCs Delete all ems records for Fugitive component SCCs Jan 2016, OKDEQ run (All Area SCCs EXCEPT Pneu. Dev. ) (Point Source Subtraction occurs) Staging Table STEP 2: Add HAPS and VOC emissions from malfunctioning pneumatic devices to Fugitives: All Process SCCs Final Staging Table 24

Procedure Flowchart: Step 3 Dec 2015, NEI run Dec 2015, Malfunctions run (All Area SCCs) (For Pneu. Dev. ONLY) Staging Table Delete all records for criteria emissions (except VOC from pneu. dev. ) Update malfun. pneu. dev. BTEX ems to new Fugitive - All Process SCCs Aggregate HAP Fugitive component SCC ems, then Append into new Fugitive - All Process SCCs Append malfun. pneu. dev. VOC ems to new Fugitive - All Process SCCs Jan 2016, OKDEQ run (All Area SCCs EXCEPT Pneu. Dev. ) (Point Source Subtraction occurs) Delete all ems records for Fugitive component SCCs Staging Table Delete all records for HAP ems Aggregate criteria Fugitive component SCC ems, then Append into new Fugitive - All Process SCCs Delete all ems records for Fugitive component SCCs Final Staging Table STEP 3: Accounts for CRITERIA emissions, and includes point source subtraction Append all records for criteria ems (except for Fugitives); Update Fugitive VOC emissions 25

Procedures: Step 3 • Same process as Step 1 (for criteria pollutants) • Aggregate fugitive component emissions by well type, and append into new Fugitives: All Process SCCs Gas Wells CBM Wells Oil Wells Connectors 2310021501 2310023511 2310011501 Flanges 2310021502 2310023512 2310011502 Open Ended Lines 2310021503 2310023513 2310011503 Valves 2310021505 2310023515 2310011505 Other 2310021506 2310023516 N/A ↓ ↓ All Process 2310021509 2310023509 2310011500 26

Procedure Flowchart: Step 4 • After filtering out facilities in our inventory that had already reported as point sources, we aggregated the remaining point emissions by SCC type (by pollutant and well type) Final Staging Table OKDEQ Emission Inventory Filter out ems from facilities already reported as point sources Aggregate point ems for related point SCCs Update aggregated Point ems to Area SCCs using Point to Nonpoint Crosswalk STEP 4: Aggregate point emissions from OKDEQ point inventory 27

Procedures: Step 4 Aggregate Point Emissions, Tanks List of Point SCC emissions aggregated in preparation for mapping to tank area SCCs. Count Sum. Of. Ems (TPY) Point SCC Point. SCC_L 4 12, 430 9, 551. 16 40400311 Fixed Roof Tank, Condensate, working+breathing+flashing losses 7, 934 7, 168. 12 31000205 Flares 1, 598 5, 406. 48 40400300 Fixed Roof Tank: Flashing Loss 1, 528 3, 808. 43 31000160 Flares 1, 607 1, 156. 84 40400302 Fixed Roof Tank: Working Loss 1, 396 1, 091. 76 40400312 Fixed Roof Tank, Crude Oil, working+breathing+flashing losses 1, 616 940. 22 40400301 Fixed Roof Tank: Breathing Loss 1, 269 413. 14 31000215 Flares Combusting Gases : 1000 BTU/scf 66 51. 37 30600903 Natural Gas 85 40. 76 40400306 External Floating Roof Tank: Withdrawal Loss 84 35. 11 30600904 Process Gas 1 2. 67 40400313 Fixed Roof Tank, Lube Oil, working+breathing+flashing losses 28

Procedure Flowchart: Step 4 • Once point emissions were aggregated into a chosen group, they were allocated to gas, oil, and CBM wells in each county using well-type ratios Final Staging Table OKDEQ Emission Inventory Filter out ems from facilities already reported as point sources Aggregate point ems for related point SCCs Update aggregated Point ems to Area SCCs using Point to Nonpoint Crosswalk STEP 4: Add point emissions to Area SCCs 29

Procedures: Step 4, Tanks • Allocate point emissions by well type using well type ratios (HPDI county averages for the permitted wellheads). 40400311 VOC Point Ems for County-X Ratio of Point Gas Wells in County-X 40400311 VOC Point Ems at Gas Wells in County-X 40400311 VOC Point Ems for County-X Ratio of Point CBM Wells in County-X 40400311 VOC Point Ems at CBM Wells in County-X 40400311 VOC Point Ems for County-X Ratio of Point Oil Wells in County-X 40400311 VOC Point Ems at Oil Wells in County-X 30

Procedures: Step 4, Tanks • Using a point to nonpoint SCC crosswalk, we added the well-type allocated point emissions to a related nonpoint SCC that was present in the Oil & Gas Tool 40400311 VOC Point Ems at Gas Wells in County-X Add To 2310021010 (VOC Ems at Condensate Tanks at Gas Wells in County-X) 40400311 VOC Point Ems at CBM Wells in County-X Add To 2310023010 (VOC Ems at Condensate Tanks at CBM Wells in County-X) 40400311 VOC Point Ems at Oil Wells in County-X Add To 2310010200 (VOC Ems at Crude Oil Tanks in County-X) 31

Point to Nonpoint Crosswalk Point_SCC Gas Wells, Nonpoint SCC CBM Wells, Nonpoint SCC Oil Wells, Nonpoint SCC Other, Nonpoint SCC Pollutants 40400311 Tanks & Flares 2310021010 Condensate Tanks 2310023010 Condensate Tanks 2310010200 Crude Oil Tanks N/A VOC, NOx , CO, SOx , PM 10 , PM 2. 5 20200253 4 -Cycle Rich 2310021302 4 -cycle Rich 2310023302 4 -cycle Rich 2310000330 Artificial Lift N/A VOC, NOx , CO, SOx , PM 10 , PM 2. 5 20200254 4 -cycle Lean 2310021202 4 -cycle Lean 2310023202 4 -cycle Lean 2310000330 Artificial Lift N/A VOC, NOx , CO, SOx , PM 10 , PM 2. 5 20200252 2 -cycle Lean 2310021102 2 -cycle Lean 2310023102 2 -cycle Lean 2310000330 Artificial Lift N/A VOC, NOx , CO, SOx , PM 10 , PM 2. 5 2310021302 4 -cycle Rich 2310023202 4 -cycle Lean 2310000330 Artificial Lift N/A VOC, NOx , CO, SOx , PM 10 , PM 2. 5 20200202 Unspecified Engines These are 50 -499 HP Engine Area SCCs. In reality, some of the engines we inventory will have HP outside of this range. 32

Point to Nonpoint Crosswalk Cont. Point_SCC Gas Wells, Nonpoint SCC CBM Wells, Nonpoint SCC Oil Wells, Nonpoint SCC Other, Nonpoint SCC Pollutants 31000220 Fugitives 2310021509 Fugitives All Process 2310023509 Fugitives All Process 2310011500 Fugitives N/A VOC 40400250 Loading Racks 2310021030 Railcar Loading: Condensate 2310023030 Railcar Loading: Condensate 2310011201 Railcar Loading: Crude Oil N/A VOC 31000404 Process Heaters 2310021100 Gas Well Heaters 23100 CBM Well Heaters 2310010100 Heater Treaters N/A VOC 40400315 Tanks, Produced Water N/A N/A 2310000550 Produced Water VOC 31000227 Dehydrators 2310021400 Dehydrators N/A N/A VOC, NOx , CO, SOx , PM 10 , PM 2. 5 33

Procedure Flowchart: Step 4 Dec 2015, NEI run Dec 2015, Malfunctions run (All Area SCCs) (For Pneu. Dev. ONLY) Staging Table Delete all records for criteria emissions (except VOC from pneu. dev. ) Update malfun. pneu. dev. BTEX ems to new Fugitive - All Process SCCs Delete all records for HAP ems Aggregate HAP Fugitive component SCC ems, then Append into new Fugitive - All Process SCCs Append malfun. pneu. dev. VOC ems to new Fugitive - All Process SCCs Aggregate criteria Fugitive component SCC ems, then Append into new Fugitive - All Process SCCs Delete all ems records for Fugitive component SCCs (All Area SCCs EXCEPT Pneu. Dev. ) (Point Source Subtraction occurs) Staging Table Delete all ems records for Fugitive component SCCs Final Staging Table OKDEQ Emission Inventory Jan 2016, OKDEQ run Filter out ems from facilities already reported as point sources Aggregate point ems for related point SCCs Append all records for criteria ems (except for Fugitives); Update Fugitive VOC emissions Update aggregated Point ems to Area SCCs using Point to Nonpoint Crosswalk 34

Final Comparison Analysis NOx (tons) VOC (tons) Dec 2015, No Change Jan 2016, NEI+Malfunctions Point Emissions FINAL 42, 166 40, 276 16, 190 56, 232 147, 984 122, 533 122, 603 38, 836 163, 037 Dec 2015, No Change Jan 2016, NEI+Malfunctions Point Emissions FINAL The Tool version released on December 4, 2015 (includes HPDI activity data and subpart W data). Point subtraction occurs. No input from OKDEQ The Tool version Total emissions released on December from OKDEQ 4, 2015 (includes HPDI aggregated point activity data and source SCCs. subpart W data). Point subtraction occurs. Total emissions from the final staging table of the flow chart procedures. Includes OKDEQ gas composition data and OIPA data for both normal and malfunction pneumatic devices. 35

Comparison of Emissions per Well (Nonpoint vs. Point) Total Number of Wells (Oil, Gas, & CBM Combined) NOX (tons) VOC (tons) NOX (tons/well) VOC (tons/well) Dec 2015 No Change Run (The 2014 O&G Tool w/o Modifications) Oklahoma Permitted Wells 55, 568 4, 306 Total Emissions 42, 166 147, 984 Per Well Emissions 0. 759 2. 663 16, 190 38, 836 3. 760 9. 019 36

Remaining Questions/Concerns • Source Classification Code structure Poor descriptions Crosswalk between point & nonpoint SCCs • Definition of gas well vs. oil well Reclassification of well types (affects where emissions are appearing and changes the emission factors in the tool) Point source inventory mitigates this problem 37

Two Final Questions 1. When ERG performed the Oklahoma point source activity subtractions, total VOC emissions (from production SCCs) decreased by 17. 2% and NOX emissions decreased by only 4. 5%. What is the reason for the difference? 2. For the 2014 Tool, the well classification criteria changed. Could this new approach be misclassifying some wells and their production (e. g. , condensate misclassified as crude oil)? 38

Question 1: Activity Changes and VOC vs. NOx Reductions In the baseline tool, gas well SCCs account for 74% of total production sector O&G NOX emissions. But the Oklahoma point source inventory of gas wells (843 wells) only represented 2. 7% of the total (unadjusted) gas wells in the baseline tool. In the baseline tool, oil well SCCs account for only 18% of production sector NOX emissions. Our point source inventory of total unadjusted oil wells (3, 397 wells) represented 16% of the total oil wells in the baseline tool. Test: Group NOX emissions by well type (oil, gas, and CBM). Reduce the NOX emissions in proportion to the reduction in wells due to the point source subtraction. N. B. : This exercise is simple, but inexact. It ignores the differences in basin factors as well as emissions driven by liquids production rather than well counts. 39

Question 1: Activity Changes and VOC vs. NOx Reductions NOx Emissions from Oil Well SCCs NOx Emissions from Gas Well SCCs NOx Emissions from CBM Well SCCs Sums (NOx Emissions) Un-Adjusted NOx Emissions (TPY) Un-Adjusted Well Counts OK Point Well Counts Adjusted Well Counts Prorated NOx Emissions (TPY) 7, 711 6, 684 21, 060 3, 397 17, 716 6, 487 31, 386 30, 574 31, 395 843 30, 552 30, 543 3, 069 3, 018 3, 113 66 3, 047 3, 004 42, 166 40, 276 Compare 40, 034 Looks reasonable. NOX emissions are driven by gas well numbers and gas wells are not as heavily represented in the OKDEQ point source inventory. 40

Question 2: Effects of the Change in the Well Classification Criterion • The 2014 Tool changed the method for classifying wells (as oil or gas) Result: a greater proportion of Oklahoma wells are classified as crude oil rather than gas (when compared to the method used in the 2011 Tool) We expect that this has resulted in an underestimation of VOC emissions from working, breathing, flashing, and loading losses. Why? Because condensate has a higher API gravity, more lighter hydrocarbons, higher vapor pressure, and more VOC emissions. NOX emissions will also be underestimated somewhat (we are not going to address this effect on NOX estimates today) 41

Oklahoma 2015 Combined Crude Oil and Condensate Production (thousands of barrels per day) API Gravity Report Month <=30. 0 30. 1 -40. 0 40. 1 -50. 0 >50. 0 Total Jan 15 60 143 147 97 448 Feb 15 62 148 149 94 454 Mar 15 60 151 161 100 473 Apr 15 56 148 162 96 463 May 15 57 138 155 95 446 Jun 15 52 133 157 89 432 Jul 15 51 132 155 86 423 Aug 15 52 128 159 82 421 Sep 15 50 126 155 83 414 Oct 15 52 121 155 86 414 Nov 15 50 115 153 86 404 Dec 15 46 105 147 98 396 54 132 155 91 432 Annual Average (Source: Energy Information Administration http: //www. eia. gov/petroleum/production/ ) Hydrocarbon liquids with API gravity ≥ 50 classified as condensate. 42

Question 2: Effects of the Change in the Well Classification Criterion • The EIA table showed total hydrocarbon liquids production by API gravity for 2015 (We did not find the data for 2014) Note: the EIA dataset breaks out API gravity of the liquids by state • The EIA data show 21% of the liquids produced to meet the definition of condensate The Oklahoma Corporation Commission classifies hydrocarbon liquids with ≥ 50 o. API gravity as condensate • The current version of the Tool only classifies 1. 5% of the total hydrocarbon liquids produced in Oklahoma as condensate Gas well condensate + CBM well condensate combined • Conclusion: the change in well classification likely does lead to an underestimate of VOC emissions. However, the Oklahoma point source data (48% of liquids captured by Oklahoma point sources) mitigates that problem. 43

Questions? Mark Gibbs Mark. Gibbs@deq. ok. gov 405 -702 -4179