ANGA/API Study of Methane Emissions from Fracked Natural Gas Wells

1. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDCharacterizing PivotalSources of MethaneEmissions fromUnconventional NaturalGas ProductionSummary and Analysis of API and ANGA SurveyResponsesTerri Shires and Miriam Lev-OnURS Corporation and The LEVON GroupFINAL REPORTJune 1, 2012

2. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDTable of Contents 1. Overview .............................................................................................................................. 1 1.1 Context.......................................................................................................................... 1 1.2 Introduction to the API/ANGA Survey ........................................................................ 3 2. Well Data ............................................................................................................................. 4 2.1 National Gas Well Counts ............................................................................................ 4 2.2 Gas Well Completions .................................................................................................. 6 2.3 Data Limitations Concerning Wells ............................................................................. 9 3. Gas Well Liquids Unloading ............................................................................................. 11 4. Hydraulic Fracturing and Re-fracturing (Workovers) ....................................................... 15 4.1 API/ANGA Survey ..................................................................................................... 15 4.2 WRAP Survey ............................................................................................................ 19 4.3 Impact of Completions and Re-fracture Rate Assumptions ....................................... 19 4.4 Completion and Re-fracture Emission Factor ............................................................ 20 4.5 Data Limitations for Completion and Re-fracture Emissions .................................... 22 5. Other Surveyed Information .............................................................................................. 23 5.1 Centrifugal Compressors ............................................................................................ 23 5.2 Pneumatic Controllers ................................................................................................ 24 6. Conclusions ........................................................................................................................ 26 7. References .......................................................................................................................... 28Appendix A. API/ANGA Survey Forms ........................................................................................ 30Appendix B. ANGA/API Well Survey Information ....................................................................... 35Appendix C. Emission Estimates for Gas Well Liquids Unloading ............................................. 39Summary and Analysis of API and ANGA Survey Responses i

3. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDTables ES-1. Emission Comparison between EPA and Industry Data ......................................... iv 1. API/ANGA Survey – Summary of Gas Well Counts by Type and NEMS Region ..... 5 2. API/ANGA Survey – Additional Details on Gas Well Counts ................................... 7 3. API/ANGA Survey – Summary of Liquids Unloading Data ...................................... 8 4. API/ANGA Survey –Liquids Unloading Emissions Comparison ............................... 8 5. API/ANGA Survey – Summary of Gas Well Workovers with Hydraulic Fracturing in 2010 and First Half of 2011 by NEMS Region And Well Type (First Survey Data Request Phase) ........................................................................................................ 12 6. API/ANGA Survey – Summary of 2010 Gas Well Workovers on Unconventional Wells by AAPG Basin and NEMS Region (Second Survey Data Request Phase) .... 13 7. WRAP Survey – Summary of Gas Well Workovers by AAPG Basin for the Rocky Mountain Region, 2006 Data ................................................................................... 16 8. API/ANGA Survey –Gas Well Workover Emissions Comparison ........................... 18 9. WRAP Survey – Summary of Completion Emissions for the Rocky Mountain Region, 2006 Data ................................................................................................... 19 10. ANGA Survey – Summary of Completion Emissions .............................................. 21 11. API/ANGA Survey –Pneumatic Controller Counts .................................................. 22 12. Pneumatic Controller Emission Comparison – Production Operations ..................... 24 13. Pneumatic Controller Emission Comparison – Production Operations ..................... 25 14. Emission Comparison between EPA and Industry Data ........................................... 26 B-1. ANGA/API Survey – Summary of Gas Well Counts by Type and NEMS Region . 36 B-2. ANGA/API Survey – Additional Details on Gas Well Counts ............................... 37 C-1. Liquids Unloading for Conventional Gas Wells without Plunger Lifts .................. 40 C-2. Liquids Unloading for Conventional Gas Wells with Plunger Lifts ....................... 41 C-3. Liquids Unloading for Unconventional Gas Wells without Plunger Lifts .............. 42 C-4. Liquids Unloading for Unconventional Gas Wells with Plunger Lifts ................... 45Figures 1. Comparison of EPA to API/ANGA Gas Well Count Data by AAPG Basin ............. 29 A.1 Survey Instructions .................................................................................................. 30 A.2 Gas Well Survey Data ............................................................................................. 31 A.3 Gas Well Workover Survey Data ............................................................................. 32 A.4 Gas Well Liquids Unloading Survey Data ............................................................... 33 A.5 Other Survey Data ................................................................................................... 34Summary and Analysis of API and ANGA Survey Responses ii

4. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDExecutive Summary This document presents the results from a collaborative effort among members of theAmerican Petroleum Institute (API) and America’s Natural Gas Alliance (ANGA) to gather dataon key natural gas production activities and equipment emission sources - includingunconventional natural gas production - that are essential to developing estimates of methaneemissions from upstream natural gas production. API and ANGA members undertook this effort as part of an overall priority to developnew and better data about natural gas production and make this information available to thepublic. This information acquired added importance in 2011, when the EPA released aninventory of U.S. greenhouse gases (GHG) emissions that substantially increased estimates ofmethane emissions from Petroleum and Natural Gas Systems. Public comments submitted byboth trade associations reflected a number of concerns – most notably that EPA’s estimates werebased on a small set of data submitted by a limited number of companies in a different context(i.e., data not developed for the purpose of estimating nationwide emissions). The API/ANGA data set (also referred to as ANGA/API) provides data on 91,000 wellsdistributed over a broad geographic area and operated by over 20 companies. This representsnearly one-fifth (18.8%) of the estimated number of total wells used in EPA’s 2010 emissionsinventory. 1 The ANGA/API data set is also more than 10 times larger than the set of wells inone of EPA’s key data sources taken from an older Natural Gas Star sample that was neverintended for developing nationwide emissions estimates. Although more and better data effortswill still be needed, API/ANGA members believe this current collaborative effort is the mostcomprehensive data set compiled for natural gas operations. As Table ES-1 demonstrates, survey results in two source categories – liquidsunloading and unconventional gas well re-fracture rates - substantially lower EPA’s estimatedemissions from natural gas production and shift Natural Gas Systems from the largestcontributor of methane emissions to the second largest (behind Enteric Fermentation, whichis a consequence of bovine digestion). 2 The right-hand column of this table shows the impact ofANGA/API data on the estimated emissions for each source category. Gas well liquidsunloading and the rate at which unconventional gas wells are re-fractured are key contributors tothe overall GHG emissions estimated by EPA in the national emissions inventory. For example,methane emissions from liquids unloading and unconventional well re-fracturing accounted for59% of EPA’s estimate for overall natural gas production sector methane emissions. Overall,API/ANGA activity data for these two source categories indicate that EPA estimates of potentialemissions from the production sector of “Natural Gas Systems” would be 50% lower if EPAwere to use ANGA/API’s larger and more recent survey results.1 EPA’s 2010 national inventory indicates a total of 484,795 gas wells (EPA, 2012).2 Table ES-2 of the 2010 national inventory (EPA, 2012).Summary and Analysis of API and ANGA Survey Responses iii

5. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED TABLE ES-1. EMISSION COMPARISON BETWEEN EPA AND INDUSTRY DATA Impact on Source Source Category Category EPA API/ANGA Emissions Metric tons of CH4 % of EPA Metric tons of CH4 % of API & ANGA - EPA Emissions Revised EPA Total Emissions % Difference in Total Emissions Gas Wells Liquids 4,501,465 * 51% 637,766 14% -86% Unloading Unconventional 712,605 * 8% 197,311 4% -72% Well Re-fracture Rates Other Production 3,585,600 41% 3,585,600 81% ** Sector Emissions Total Production 8,799,670 4,420,677 -50% Sector Emissions * EPA’s estimates are adjusted to industry standard conditions of 60 degrees F and 14.7 psia for comparison to the ANGA/API emission estimates. ** The “Other Production Sector Emissions” are comprised of over 30 different source categories detailed in Table A-129 in the Annex of the EPA’s 2012 national inventory. The “Other Production Sector Emissions” are the same values for this comparison between the EPA national inventory and the API/ANGA survey to focus the comparison on quantified differences in emission estimates for gas well liquids unloading and unconventional well re-fracture rates. As mentioned above, the differences between EPA and ANGA/API estimates hinge onthe following key differences in activity data and thus considerably impact overall emissionsfrom Natural Gas Systems: • Liquids unloading and venting. API/ANGA data showed lower average vent times as well as a lower percentage of wells with plunger lifts and wells venting to the atmosphere than EPA assumed. This is particularly significant because liquids unloading accounted for 51% of EPA’s total “Natural Gas Systems” methane emissions in the 2010 inventory. Applying emission factors based on ANGA/API data reduces the calculated emissions for this source by 86% (from 4,501,465 metric tons of CH 4 to 637,766 metric tons of CH 4 when compared on an equivalent basis) from EPA’s 2010 national GHG inventory. • Re-fracture rates for unconventional wells. API/ANGA members collected data on re- fracture rates for unconventional wells in two phases. The first phase collected data for all well types (conventional and unconventional), while the second phase targeted unconventional gas wells. Both phases of the survey data show significantly lower rates of well re-fracturing than the 10% assumption used by EPA. As discussed in detail in this report, the re-fracture rate varied from 0.7% to 2.3%. The second phase of the survey gathered data from only unconventional well activity and using the re-fracture rate data from this second phase of the ANGA/API survey reduces the national emission estimateSummary and Analysis of API and ANGA Survey Responses iv

6. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED for this source category by 72%, - from 712,605 metric tons of CH 4 to 197,311 metric tons of CH 4 when compared on an equivalent basis. This report also discusses an important related concern that the government lacks a singlecoordinated and cohesive estimate of well completions and well counts. Although the 2010national GHG inventory appears to under-represent the number of well completions according tothe numbers reported through both the API/ANGA data and IHS CERA, differences in nationalwell data reporting systems make it difficult to accurately investigate well completiondifferences with any certainty. The EPA inventory, which uses data from HPDI, and the EnergyInformation Administration (in addition to privately sourced data) all report different well countsthat do not consistently distinguish between conventional and unconventional wells. Without aconsistent measure for the quantity and type of wells, it is difficult to be confident of theaccuracy of the number of wells that are completed annually, let alone the amount of emissionsfrom them. Natural gas producers strongly believe that the effects of any possible under-representation of well completions will be offset by a more realistic emission factor for the rateof emissions per well. This survey also collected data on centrifugal compressors and pneumatic controllers.While the sample sizes are too small to make strong conclusions, the results discussed in thebody of the report indicate that further research is necessary to accurately account for thedifferent types of equipment in this area (e.g., wet vs. dry seal centrifugal compressors and “highbleed,” “low bleed,” and “intermittent bleed” pneumatic controllers). As government and industry move forward in addressing emissions from unconventionalgas operations, three key points are worth noting: • In addition to the voluntary measures undertaken by industry, more data will become available in the future. Emission reporting requirements under Subpart W of the national Greenhouse Gas Reporting Program (GHGRP) went into effect January 1, 2011 with the first reporting due in the fall of 2012. As implementation of the GHGRP progresses from year to year, the natural gas industry will report more complete and more accurate data. If EPA makes use of the data submitted and transparently communicates their analyses, ANGA/API members believe this will increase public confidence in the emissions estimated for key emission source categories of the Natural Gas Systems sector. • Industry has a continuous commitment to improvement. It is clear that companies are not waiting for regulatory mandates or incentives to upgrade equipment, or to alter practices like venting and flaring in favor of capturing methane where practical. Instead, operators are seizing opportunities to reduce the potential environmental impacts of their operations. Industry is therefore confident that additional, systematic collection of production sector activity data will not only help target areas for future reductions but also demonstrate significant voluntary progress toward continually ‘greener’ operations. • Members of industry participating in this survey are committed to providing information about the new and fast-changing area of unconventional oil and gas operations. API and ANGA members look forward to working with the EPA to revise current assessment methodologies as well as promote the accurate and defensible uses of existing data sources.Summary and Analysis of API and ANGA Survey Responses v

7. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED 1. Overview The accuracy of GHG emission estimates from unconventional natural gas productionhas become a matter of increasing public debate due in part to limited data, variability in thecomplex calculation methodologies, and assumptions used to approximate emissions wheremeasurements in large part are sparse to date. Virtually all operators have comprehensivemethane mitigation strategies; however, beyond the requirements of the EnvironmentalProtection Agency’s (EPA) Mandatory Reporting Rule or incentives of programs like the EPA’sNatural Gas Star program, data is often not gathered in a unified way that facilitates comparisonamong companies. In an attempt to provide additional data and identify uncertainty in existing data sets, theAmerican Petroleum Institute (API) and America’s Natural Gas Alliance (ANGA) began a jointstudy on methane (CH 4 ) emissions from unconventional gas operations in July 2011. The firstpart of this section offers context to the decision to conduct this survey, while the second offers abrief introduction to the survey itself. 1.1 Context Shale gas will undoubtedly play a key role in America’s energy future and thereforeadditional information must be collected to quantify the methane emissions from bothconventional and unconventional natural gas production. Meaningful, publicly available data isa priority, especially in light of EPA’s 2011 revision of its calculation methodology for NaturalGas Systems in the 2009 national inventory (EPA, 2011b). (EPA added two new sources forunconventional gas well completions and workovers, and also significantly revised its estimatesfor liquids unloading and made adjustments to other source categories.) These changessubstantially increased EPA’s estimated GHG emissions for the production sector of the NaturalGas Systems by 204%. Industry was alarmed by the upward adjustment, especially since previous EPA estimateshad been based on a 1996 report prepared by the EPA and GRI – and did not take into accountthe considerable improvements in equipment and industry practice that have occurred in thefifteen years between 1996 and 2011 (GRI, 1996). An EPA technical note to the 2009 inventory attributed the changes to adjustments incalculation methods for existing sources, including gas well liquids unloading, condensatestorage tanks, and centrifugal compressor seals. EPA also added two new sources not previouslyincluded in its inventories, namely unconventional gas well completions and workovers (re-completions) (EPA, 2011e). Industry did not have an adequate opportunity to examine EPA’s rationale for the newemissions factor prior to its initial release. Unlike changes in regulatory requirements, EPA isnot required to initiate a formal comment process for changes in methodologies like emissionfactors and calculations methods in the national GHG inventory. As such, EPA is not compelledto incorporate or consider input provided by stakeholders and experts. Indeed, changes tomethodologies are often made without the benefit of dialogue or expert review. Although EPAfurther acknowledged in the 2010 inventory (released in 2012), that their natural gas calculationsneeded work, their practice is to continue using the same numbers until adjusted estimates haveSummary and Analysis of API and ANGA Survey Responses 1

8. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDbeen made. It is important to note that EPA has indicated a willingness to engage and discussthis matter with some members of industry; however, no time frame has yet been determined forthis discussion. Under the best of circumstances, EPA had remarkably little information to draw on indetermining their new emission factor. Input from industry on this topic was not directlysolicited. Specific guidance also did not exist on the international level, nor was it availablefrom other national regulators. A review of the Intergovernmental Panel on Climate Change(IPCC) and other inventories submitted to the United Nations Framework Convention onClimate Change (UNFCCC) indicate that the U.S. is currently the only country to date todifferentiate between conventional and unconventional natural gas production. Regulators,academics, and environmentalists around the world therefore considered the new estimatedemission factor as an unprecedented development in a controversial issue. Widespread criticism of the figures revealed problematic methodology and lessjustification for the underlying numbers than originally anticipated. In a paper entitledMismeasuring Methane, the well-respected energy consultancy IHS CERA succinctly detailedseveral concerns about the revisions – most notably that EPA’s new estimate was based on onlyfour (4) data points that natural gas well operators had submitted voluntarily under the NaturalGas Star Program, which highlights emissions reductions. Together, the four data points coverapproximately 8,880 wells – or roughly 2% of those wells covered in the EPA’s nationalgreenhouse gas inventory. Those numbers, which were submitted in the context of showcasingachieved emissions reductions and not to estimate emissions, were then extrapolated to over488,000 wells in the 2009 emissions inventory (IHS CERA, 2011). With an emerging topic like shale energy development, however, the impact of EPA’srevised estimates was enormous. Emission estimates from production using EPA’s figures wereused to question the overall environmental benefits of natural gas. They were cited widely byunconventional gas opponents - many of whom used the new figures selectively and withoutcaveats like “estimated” to argue against further development of shale energy resources. Forexample, an article published by ProPublica cited the revised EPA emission factors as “newresearch” which “casts doubt” on whether natural gas contributes lower GHG emissions thanother fossil fuels (Lustgarten, 2011). Many of these studies – e.g., the work of Howarth et al.were widely reported in the popular press (Zellers, 2011) with little attention to the quality ofanalysis behind their conclusions. Notably, other authors using more robust and defensible scientific methodologies arguedthat - even with undoubtedly high emissions estimates - natural gas still possessed a lifecycleadvantage when its comparative efficiency in electricity generation was taken into account. Forexample, a study by Argonne National Laboratory utilizing the same EPA data sourcesconcluded that taking into account power plant efficiencies, electricity from natural gas showssignificant life-cycle GHG benefits over coal power plants (Burnham, 2011). Unfortunately, thecomplex technical arguments in these studies generated considerably less media and publicattention. It is important to understand that the ongoing debate about the accuracy of EPA’sadjusted emission factor as contained in the 2009 inventory did not keep these numbers frombeing used in a series of rules that have wide ranging ramifications on national natural gaspolicies both in the United States and globally. Many countries considering shale energySummary and Analysis of API and ANGA Survey Responses 2

9. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDdevelopment remain bound by the emissions reduction targets in the Kyoto Protocol and theirregulatory discussions reflect greenhouse gas concerns. In addition to the very real risk thatother countries could adopt the emission factor before the EPA can refine its calculations, thepossibility of higher emissions (even if only on paper) might deter other nations from developingtheir own unconventional energy resources. By the summer of 2011, it was clear to ANGA/API members (also referred to asAPI/ANGA members) that gathering additional data about actual emissions and points ofuncertainty during unconventional gas production was essential to improve GHG life cycleanalysis (LCA) of natural gas for the following reasons: 1) to focus the discussion of emissionsfrom natural gas production around real data; 2) to promote future measurement and mitigationof emissions from natural gas production; and 3) to contribute to improving the emissionestimation methods used by EPA for the natural gas sector in their annual national GHGinventory. 1.2 Introduction to the API/ANGA Survey API and ANGA members uniformly believed that EPA’s current GHG emissionsestimates for the natural gas production sector were overstated due to erroneous activity data inseveral key areas - including liquids unloading, well re-fracturing, centrifugal compressors, andpneumatic controllers. Members therefore worked cooperatively to gather information throughtwo data requests tailored to focus on these areas and reasonably accessible information aboutindustry activities and practices. Specifically, information was requested on gas well types, gaswell venting/flaring from completions, workovers, and liquids unloading, and the use ofcentrifugal compressor and pneumatic controllers. The actual data requests sent to members can be found in Appendix A, and Appendix Bprovides more detailed data from the ANGA/API well survey information. Survey results and summaries of observations, including comparisons to EPA’s emissionestimation methods, are provided in the following sections.Summary and Analysis of API and ANGA Survey Responses 3

10. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED 2. Well Data This section examines well data gathered by API and ANGA members. Overall,ANGA/API’s survey effort gathered activity data from over 20 companies covering nearly91,000 wells and 19 of the 21 American Association of Petroleum Geologists (AAPG) basins 3containing over 1% of the total well count in EPA’s database of gas wells. Members believe thatthe API/ANGA survey represents the most comprehensive data set ever compiled for natural gasoperations and, as such, provides a much more accurate picture of operations and emissions. Information to characterize natural gas producing wells was collected by survey in twoparts: • The first part of the survey requested high-level information on the total number of operating gas wells, the number of gas well completions, and the number of gas well workovers with hydraulic fracturing. Data on over 91,000 wells was collected primarily for 2010, with some information provided for the first half of 2011. • The second part of the survey requested more detailed well information about key activities. The well information collected through the two surveys is provided in Appendix B. Section 2.1 looks at overall natural gas well counts, Section 2.2 examines completiondata from ANGA/API members, and Section 2.3 briefly identifies several unresolved issuesconcerning well counts and classifications that could benefit from future analysis forexamination. For the purposes of this report, unconventional wells are considered to be shale gaswells, coal bed wells, and tight sand wells which must be fractured to produce economically.2.1 National Gas Well Counts To provide context for the information collected by API and ANGA, comparisons weremade to information about national gas wells from EPA and the U.S. Energy InformationAdministration (EIA). Unfortunately, the government lacks a single coordinated and cohesiveset of estimates for gas wells. Industry grew concerned when it became apparent that significant discrepancies existedamong different sources of national gas well data. The EPA inventory, the EIA, and IHS allreported different well counts that do not consistently distinguish between key areas likeconventional and unconventional wells. Furthermore, there does not appear to be a singletechnical description for classifying wells that is widely accepted. Without consistent measuresand definitions for the quantity and type of wells, it is difficult to reach agreement on the numberof unconventional wells completed annually - let alone their emissions.3 Basins are defined by the American Association of Petroleum Geologists (AAPG) AAPG–CSD GeologicProvinces Code Map: AAPG Bulletin, Prepared by Richard F. Meyer, Laure G. Wallace, and Fred J. Wagner, Jr.,Volume 75, Number 10 (October 1991) and the Alaska Geological Province Boundary Map, Compiled by theAmerican Association of Petroleum Geologists Committee on Statistics of Drilling in Cooperation with the USGS,1978.Summary and Analysis of API and ANGA Survey Responses 4

11. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED Both the EIA data and the EPA data accompanying the national GHG inventory lacksufficient detail for well classifications to provide a basis for helpful comparison with the surveydata reported here. Instead, national well data developed as part of mandatory emissionsreporting is used for comparison because it has the most appropriate level of detail in wellcategories (EPA, 2011d). In EPA’s database gas well count (EPA, 2011d), 21 of the AAPG basins each have morethan 1% of the total well count. The API/ANGA survey has wells from 19 of those 21 basins. Interms of wells represented by these basins, 92% of the total EPA database well count isaccounted for by wells in those 21 basins, while 95% of the ANGA/API surveyed gas wells areaccounted for by those 21 basins. These results are summarized in Table 1 and illustrated inFigure 1. This indicates that the API/ANGA survey results have good representation for thebasins with the largest numbers of wells nationally. TABLE 1. COMPARISON OF GAS WELL COUNT DATA BY AAPG BASIN: SUMMARY STATISTICS EPA Database API/ANGA Survey ANGA/API as a Gas Well Data % of EPA Count* Total number of U.S. gas wells 355,082 gas wells 91,028 gas wells 26% Number of significant AAPG 21 basins Data on wells in 19 of 90% basins** those 21 basins Number of wells in significant AAPG 325,338 wells 86,759 wells 27% basins % of total wells in significant AAPG 92% 95% basins* EPA’s database gas well count (EPA, 2011d) differs from the well count provided in EPA’s 2010 nationalinventory, but provides more detail on the types of wells. Additional details are provided in Appendix B.** Significant basins are defined as basins with more than 1% of the total national gas wells. As shown in Figure 1, the API/ANGA survey results more heavily represent gas wells inspecific AAPG basins when compared to EPA’s basin-level well counts (EPA, 2011c). Unlikethe EPA data, the ANGA/API data is more heavily influenced by AAPG 160 and 160A. AAPGbasins 360, 230, and 580 are important for both data sets. The smaller data set provided by EPA (2011d) may not include all of the Marcellus shalewells (particularly in Pennsylvania), and the well classification system used in this smaller dataset could probably be made more rigorous. Although this comparison may not show a perfectdistributional match for the basin by basin distribution of the API/ANGA survey data presentedhere, it does not change the fundamental conclusion of the ANGA/API survey since this data setdoes cover 90% of the basins and 27% of the national gas well count for the significant basins asreported by EPA (EPA, 2011d). The data discussed in this report provides substantial newinformation for understanding the emissions from Natural Gas Systems and offers a compellingjustification for re-examining the current emission estimates for unconventional gas wells. Appendix B contains more detail about the industry well data sample compared to theoverall data maintained by the government. Unless otherwise noted, further statisticalcomparisons of well data throughout this paper are done with reference to the EPA data becauseit was the only one which effectively parsed the data by well type (EPA, 2011d).Summary and Analysis of API and ANGA Survey Responses 5

12. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED FIGURE 1. COMPARISON OF EPA TO API/ANGA GAS WELL COUNT DATA BY AAPG BASIN 2.2 Gas Well Completions Acknowledging the somewhat different time periods covered, the API/ANGA surveydata represents 57.5% of the national data for tight gas well completions and 44.5% of shale gaswell completions, but only 7.5% of the national conventional well completions and 1.5% of coal-bed methane well completions. About one-third of the surveyed well completions (2,205) couldnot be classified into the well types requested (i.e., tight, shale, or coal-bed methane). The surveyresults for well completions are provided in Table 2 and compared to national data provided toANGA by IHS. 4 EPAs 2010 inventory showed 4,169 gas well completions with hydraulic fracturing(EPA, 2012, Table A-122); however, EPA does not provide a breakout of completions by welltype (shale gas, tight gas or coal-bed methane). In comparing the EPA 2010 count of gas wellcompletions with hydraulic fracturing (4,169 completions) to both the survey results and data4 Data provided in e-mail from Mary Barcella (IHS) to Sara Banaszak (ANGA) on August 29,2011. Data werepulled from current IHS well database and represent calendar year 2009 (2010 data are not yet available).Summary and Analysis of API and ANGA Survey Responses 6

13. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDprovided by IHS, it seems that EPA’s national GHG inventory underestimates the number ofwell completions. Even accounting for the difference in time periods (2010 for EPA comparedto 2010/2011 data from the ANGA/API survey), the national inventory appears to under-represent the number of well completions. TABLE 2. API/ANGA SURVEY – SUMMARY OF GAS WELL COMPLETIONS BY NEMS REGION AND WELL TYPE* (FIRST SURVEY DATA REQUEST PHASE) Conventional Coal-bed RegionalNEMS Region Wells Shale Methane Tight Unspecified Total API/ANGA Survey Data Gas Well CompletionsNortheast 2 291 3 67 126 489Gulf Coast 81 588 - 763 374 1,806Mid-Continent 22 734 - 375 270 1,401Southwest 425 442 - 346 310 1,523Rocky Mountain 10 30 977 1,017Unspecified - - - - 1,125 1,125Survey TOTAL 540 2,055 33 2,528 2,205 7,361% of Survey Total 7.3% 27.9% 0.4% 34.3% 30.0% 2010 IHS Gas Well Completions IHS Total 2010 National 7,178 4,620 2,254 4,400 18,452 Well Completions (from IHS)1 38.9% 25.0% 12.2% 23.8%API/ANGA as % ofIHS National Well 7.5% 44.5% 1.5% 57.5%Counts * ANGA/API survey data represents well counts current for calendar year 2010 or the first half of 2011. ** EPA’s national GHG inventory does not designate gas wells by classifications of “shale”, “coal bed methane” or “tight”. As shown in Table 3, the ANGA/API survey noted 7,361 gas well completions for 2010and the first half of 2011. This is equivalent to approximately 40% of the gas well completionsreported by IHS for 2010. Although EPA’s 2010 national GHG inventory appears to under-represent the number of gas well completions according to the numbers reported through boththe API/ANGA data and the IHS, differences in national well data reporting systems make itdifficult to accurately investigate well completion differences with certainty. The EPAinventory, which uses data from HPDI, and the Energy Information Administration (in additionto privately sourced data) - all of which report different well counts that do not consistentlydistinguish between conventional and unconventional wells. Without a consistent measure forthe quantity and type of wells, it is difficult to be confident of the accuracy of how many wellsare completed annually, let alone to estimate their emissions. Industry strongly believes that theeffects of any current under-representation of well completions will be offset by a more realisticemission factor for the rate of emissions per well.Summary and Analysis of API and ANGA Survey Responses 7

14. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED TABLE 3. SUMMARY OF GAS WELL COMPLETIONS DATA (FIRST SURVEY DATA REQUEST PHASE) # Completions for Gas Wells # Completions without for Gas Wells hydraulic with hydraulic Total fracturing fracturing Completions 2010 National Well Completions (from EPA; EPA 2012) 702 4,169 4,871 % of National Total 14% 86% API/ANGA Survey Well Completions 540 6,821 7,361 % of National Total 7% 93% Well Completions from IHS 7,178 11,274 18,452 % of National Total 39% 61% Table 4 provides detailed data for well completions from the ANGA/API survey. Fromthe survey, 94% of gas well completions in 2010 and the first half of 2011, were conducted onwells with hydraulic fracturing. About one-half of all gas well completions for this time periodwere for tight wells, and about one-half of all gas well completions were for vertical wells withhydraulic fracturing. Any differences in totals between Tables 2, 3 and 4 are because thesetables were derived from the two different data requests sent to member companies as describedpreviously in the introduction to Section 2. TABLE 4. API/ANGA SURVEY – ADDITIONAL DETAILS ON GAS WELL COMPLETIONS (SECOND SURVEY DATA REQUEST PHASE ) # Completions for Gas Wells with hydraulic Gas Wells without Completions fracturing (HF) hydraulic fracturing Total % of % of # Vertical # Horizontal Total Wells Wells wells well Wells with # without completions completions with HF HF Completions HFTOTAL 315 57 372 69% 164 31% 536ConventionalTOTAL Shale 317 1,863 2,180 99% 30 1% 2,210TOTAL Tight 2,054 368 2,422 96% 106 4% 2,528TOTAL Coal Bed 27 3 30 91% 3 9% 33MethaneTOTAL OVERALL 2,713 2,291 5,004 94% 303 6% 5,307 The following points summarize survey information provided in Tables 2, 3 and 4. Thesetables represent a snapshot of well activity data during this time.Summary and Analysis of API and ANGA Survey Responses 8

15. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED • Overall, the survey showed 94% of the 5,307 wells reported in the API/ANGA data set as completed in 2010 and the first half of 2011 used hydraulic fracturing. • 536 conventional gas wells were completed in 2010 and the first half 2011. ◦ 59% were vertical wells with hydraulic fracturing, ◦ 11% were horizontal wells with hydraulic fracturing, and ◦ 31% were wells without hydraulic fracturing. • 2,210 shale gas wells were completed in 2010 and the first half 2011. ◦ 14% were vertical wells with hydraulic fracturing, ◦ 84% were horizontal wells with hydraulic fracturing, and ◦ 1% were wells without hydraulic fracturing. • 2,528 tight gas wells were completed in 2010 and the first half 2011. ◦ 81% were vertical wells with hydraulic fracturing, ◦ 15% were horizontal wells with hydraulic fracturing, and ◦ 4% were wells without hydraulic fracturing. • 33 coal-bed methane wells were completed in 2010 and the first half 2011. ◦ 82% were vertical wells with hydraulic fracturing, ◦ 9% were horizontal wells with hydraulic fracturing, and ◦ 9% were wells without hydraulic fracturing. 2.3 Data Limitations Concerning Wells In response to follow-up questions on well data, EPA indicated that they classified gaswell formations into four types (conventional, tight, shale, and coal-bed) (EPA, 2011d). Whendeveloping the gas well classifications, EPA applied their judgment where data were notavailable in the database. ANGA and API are interested in using the well database compiled byIHS or a similar database, to more completely classify gas wells at some point in the future. TheAPI/ANGA survey did not specifically define conventional wells for collecting the well datapresented in this section, leaving the respondents to determine the classification of wells basedon their knowledge of the well characteristics or state classifications. As such, this wellclassification may vary somewhat according to the respondent’s classification of wells. It should be noted that there is not a generally accepted definition for “gas wells.”Producers might be producing from several zones in the same formation, and different statesdefine “gas” or “oil” wells differently due to the historical structure of royalties and revenues.There is also no commonly used definition of “conventional” gas wells. Thus, differentdefinitions of these terms may have produced inconsistency in the classification of wells betweengas and oil, and conventional and unconventional for the surveyed results, as well as for the EPAand EIA national data. For the purposes of this report, unconventional wells are considered to beSummary and Analysis of API and ANGA Survey Responses 9

16. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDshale gas wells, coal bed wells, and tight sand wells which must be fractured to produceeconomically.Summary and Analysis of API and ANGA Survey Responses 10

17. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED 3. Gas Well Liquids Unloading Gas well clean ups also known as liquids unloading accounts for 51% of total CH 4emissions from the natural gas production sector in EPA’s national GHG inventory (EPA,2012). 5 This was a considerable increase from the 6% of CH 4 emissions that liquids unloadingrepresented in the 2008 inventory. The accuracy of assumptions regarding this activity wastherefore a major concern to API/ANGA members. As the name indicates, liquids unloading is a technique to remove water and other liquidsfrom the wellbore so as to improve the flow of natural gas in conventional wells andunconventional wells. In EPA’s national inventory, emissions from gas well liquids unloading are based on thefollowing assumptions: • 41.3% of conventional wells require liquids unloading. • 150,000 plunger lifts are in service, which equates to 42% of gas wells. • The average gas well is blown down to the atmosphere 38.73 times per year. • The average casing diameter is 5 inches. • A gas well is vented to the atmosphere for 3 hours once the liquids are cleared from the well. The ANGA/API survey gathered activity and emissions related information for gas wellliquids unloading. Information was received covering eight conventional well data sets and 26unconventional well data sets. The following information was requested: • Geographic area represented by the information provided; • Time period – data were annualized to 12 months if the information was provided for a partial year; • Number of operated gas wells represented by the information provided; • Number of gas wells with plunger lift installed; • Number of gas wells with other artificial lift (beam pump; ESP; etc.); • Total number of gas well vents; • Number of wells with and without plunger lifts that vent to the atmosphere; • Total count of gas well vents for time period with and without plunger lifts; • Average venting time for wells with and without plunger lifts; • Average daily production of venting gas wells (Mcf/day); • Average depth of venting wells (feet);5 See EPA Table A-129, of Annex 3 of the 2010 inventory report.Summary and Analysis of API and ANGA Survey Responses 11

18. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED • Average casing diameter of venting gas wells (inches); • Average tubing diameter of venting gas wells with plunger lift (inches); and • Average surface pressure - venting gas wells (psig). Table 5 summarizes the results from the API/ANGA survey and compares the results tothe assumptions EPA uses to estimate emissions for this source in the national GHG inventory. The ANGA/API data differed from EPA’s assumptions in several ways: 1) API/ANGA showed lower percentages of wells with plunger lifts; 2) API/ANGA data indicated lower percentages of wells venting to the atmosphere; 3) API/ANGA data showed lower average vent times than EPA’s numbers; and 4) Casing diameters from the API/ANGA survey were comparable to EPA’s assumption of 5 inches. TABLE 5. ANGA/API SURVEY – SUMMARY OF LIQUIDS UNLOADING DATA API/ANGA Survey Unconventional EPA Parameter Conventional Wells Wells Assumptions Number of gas wells with plunger 10% 45% 42% lifts Number of gas wells with other 25% 7% artificial lift (beam pump, ESP, etc.) Number of gas wells vented to the 11% 16% 41.3% atmosphere for liquids unloading 303.9 (all data)* # vents per well (weighted average) 33.6 38.7 32.4 (w/o outliers) ** Average venting time per vent (weighted average) With plunger lifts 0.25 hours 0.77 hours 3 hours Without plunger lifts 1.78 hours 1.48 hours Weighted Average casing diameter 4.64 inches 5.17 inches 5 inches Weighted Average tubing diameter 2.27 inches 2.43 inches Average Emission factor, Mscf/well 823 (all data)* With plunger lifts 196 14.7 (w/o outliers)** Without plunger lifts 56.4 318 Weighted average Methane 175* 1,316 emission factor, Mscf CH4/well * Includes all liquids unloading data from the ANGA/API survey ** Excluding two high data pointsSummary and Analysis of API and ANGA Survey Responses 12

19. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED When examining Table 5, it is important to note the presence of several outliers. Two data responses for operations with conventional wells reported very high frequencies of vents to the atmosphere. These data sets represent 174 gas wells with plunger lifts (out of a total 788 gas wells with plunger lifts represented by the total data set) located in the Mid-Continent region. The wells represented by these data points have plunger lifts that vent to the atmosphere for each plunger cycle. The information was confirmed by the two data respondents and is an artifact of the plunger control for these wells which results in very short venting durations (between 4 and 5 minutes) for each plunger cycle. As a result, accounting for the high frequency of plunger lift cycles for these wells results in a high average vent frequency, but still produces a lower emission factor than the EPA assumptions. Excluding these two data points, the API/ANGA survey data for the number of vents per well was comparable to EPA’s assumed frequency. Moreover, even with the high frequency of vents from these wells, the emissions are much lower than EPA’s estimates (see Table 6). TABLE 6. ANGA/API SURVEY –LIQUIDS UNLOADING EMISSIONS COMPARISON API/ANGA Survey EPA Inventory API & ANGA - EPA Estimated EPA Emission Estimated Emission Emissions, Factor, Mscf Emissions, Factor, Mscf tonnes % Difference inNEMS Region CH4/well tonnes CH 4 # wells CH 4 /well CH 4 * EmissionsNortheast 136 202,503 77,931 1,360 2,027,265 -90%Mid Continent 392 235,813 31,427 703 422,893 -44%Rocky 26,620Mountain 177 90,387 690 351,672 -74%Southwest 36 7,913 11,444 865 189,407 -96%Gulf Coast 169 101,150 31,331 2,519 1,510,259 -93% Excluded forWest Coast No data for this region 638 1,492 consistent comparison 175 (weightedTOTAL average) 637,766 179,391 4,501,465 -86% *EPA estimated emissions = # wells × EPA emission factor, converted to mass emissions based on 60 degrees F and 14.7 psia These variances among operators in ANGA/API data demonstrate the challenge of applying national emissions estimates to conditions in which there can be considerable variation in wells and operating techniques, among and even within various regions. As member companies have noted in various comments to regulators, oil and natural gas production operations vary considerably according to factors such as local geology, hydrology, and state law. EPA noted that wells equipped with plunger lifts have approximately 60% lower emissions from liquids unloading than wells without plunger lifts (EPA, 2011b). From the API/ANGA survey, an emission reduction of about 38% was observed for the unconventional Summary and Analysis of API and ANGA Survey Responses 13

20. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDwells equipped with plunger lifts compared to those without plunger lifts. However, Table 5indicates that for conventional gas wells, the average emission factor is higher for wells withplunger lifts compared to those without when the two high data points are included. Excludingthe two high data points, the emission factor for conventional wells with plunger lifts is 74%lower than the emission factor for conventional wells without plunger lifts. One reason for this discrepancy in the data may be that EPA has acknowledged that theircurrent estimation method for liquids unloading does not account for activities used to reduceCH 4 emissions by many different artificial lift methods used in industry. According to NaturalGas Star Reports, the applicable emission reductions range from 4,700 to 18,250 Mscf/yr forplunger lift systems (EPA, 2006); however, since the emission reductions are reported separatefrom the emission estimate in the national inventory, they cannot be linked back to EPAemission source categories. Emissions were calculated by applying Equation W-8 or W-9 from the EPA GHGreporting rule in 40 CFR 98 Subpart W, where Equation W-8 applies to gas wells withoutplunger lifts, and Equation W-9 applies to gas wells with plunger lifts. Appendix C summarizesthe data collected and estimated emissions. The emission results are shown in Table 6 by NEMSregion for comparison to EPA’s emission estimates. The ANGA/API survey averaged theemission factors data within each NEMS region for conventional and unconventional wellscombined. The emission results shown in Table 6 were determined by applying the API/ANGAemission factors and EPA emission factors, respectively, to the total number of wells requiringliquids unloading from the 2010 national GHG inventory. As production companies continue to collect information for EPA’s mandatory GHGreporting program, better information on liquids unloading frequency and emissions will beavailable. One area that would benefit from additional information is an investigation ofregional differences, or plunger lift control practices, in view of the high frequency of ventsobserved for two data sets containing conventional gas wells with plunger lifts in the Mid-Continent region.Key findings of the ANGA/API survey on liquids unloading are:• For all of the NEMS regions, the API/ANGA survey data resulted in lower emission estimates than EPA estimated for the 2010 national GHG inventory when compared on a consistent basis.• Overall, the change in emission factors based on data collected from the ANGA/API survey reduces estimated emissions for this source by 86% from the emissions reported in EPA’s 2010 national GHG inventory.Summary and Analysis of API and ANGA Survey Responses 14

21. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED 4. Hydraulic Fracturing and Re-fracturing (Workovers) A well workover refers to remedial operations on producing natural gas wells to try toincrease production. Starting with the 2009 inventory, EPA split the estimation of emissionsfrom producing gas wells into conventional (i.e., without hydraulic fracturing) andunconventional (i.e., with hydraulic fracturing). For workovers of wells without hydraulicfracturing, the 2009 and 2010 national inventories used emission factors of the same order ofmagnitude as the 2008 inventory (2,454 scf of CH 4 /workover). In contrast, the unconventional(with hydraulic fracturing) well workover emission factor increased by a factor of three thousand(3,000). EPA did acknowledge that the new emission factor for well workovers was based onlimited information (EPA, 2011a). Moreover, several publications including MismeasuringMethane by IHS CERA underscored the perils of extrapolating estimates using only four (4) datapoints representing approximately two percent (2%) of wells – particularly when the data wassubmitted in the context of the Natural Gas Star program, which was designed to highlightemissions reduction options (IHS CERA, 2011). Unfortunately, even if the EPA’s workoverfactor is high, it must be used in estimated emissions calculations until it is officially changed. EPA’s new emission factor is 9.175 MMscf of natural gas per re-fracture (equivalent to7.623 MMscf CH 4 /re-fracture). Additionally, EPA used this new emission factor in conjunctionwith an assumed re-fracture rate of 10% for unconventional gas well workovers each year toarrive at their GHG emission estimate for this particular category. 4.1 API/ANGA Survey The ANGA/API survey requested counts for gas well workovers or re-fractures in twoseparate phases of the survey, covering 91,028 total gas wells (Table 7 covering 2010 and firsthalf of 2011 data) and 69,034 unconventional gas wells (Table 8, 2010 data only), respectively. The first phase of the survey was part of the general well data request. Counts ofworkovers by well type (conventional, tight, shale, and coal bed methane) and by AAPG basinwere requested. The frequency of workovers was calculated by dividing the reported workoverrates by the reported total number of each type of gas well. These results are summarized inTable 7, which includes a comparison to national workover data from EPA’s annual GHGinventory. The high number of workovers in the Rocky Mountain region is discussed furtherbelow. Table 7 indicates that even for the high workover rates associated with unconventionaltight gas wells, the workover rate is much less than EPA’s assumed 10% of gas wells re-fractured each year. Based on this first phase of the survey, • The overall workover rate involving hydraulic fracturing was 1.6%. • However, many of these workovers were in a single area, AAPG-540, where workovers are known to be conducted more routinely than in the rest of the country (as described in more detail below Table 8). Excluding AAPG 540, the overall workover rate involving hydraulic fracturing was 0.7%Summary and Analysis of API and ANGA Survey Responses 15

22. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED • For all unconventional wells in Table 7, the overall workover rate involving hydraulic fracturing was 2.2%. Excluding AAPG 540, the overall workover rate involving hydraulic fracturing was 0.9%.TABLE 7. API/ANGA SURVEY – SUMMARY OF GAS WELL WORKOVERS WITH HYDRAULIC FRACTURING IN 2010 AND FIRST HALF OF 2011 BY NEMS REGION AND WELL TYPE (FIRST PHASE DATA SURVEY) Unconventional Wells Conventional Coal-bed NEMS Region Wells Shale Methane Tight Unspecified Northeast - - - - - Gulf Coast - 5 - 38 73 Mid-Continent 8 1 - 73 33 Southwest 60 25 - 8 7 Rocky Mountain 4 - 25 901 - West Coast - - - - - Unspecified - - - - 200 31 25 1,020 Survey TOTAL 72 313 1,076 % of national 0.3% 21.3% Overall Survey Total 1,461 % of national 5.6% Conventional Wells Unconventional Wells National Workover Counts (from EPA’s 2010 national 21,088 5,044 inventory) 80.7% 19.3% 26,132 Unconventional Wells Conventional Coal-bed Wells Shale Methane Tight Unspecified % Workover Rate with Hydraulic Fracturing (from ANGA/API Survey) 0.3% 0.3% 0.5% 3.0% 2.4% Tight w/out AAPG 540 0.5% Unconventional Wells 2.2% W/out AAPG 540 0.9% All Wells 1.6% All Wells w/out AAPG 540 0.7%Summary and Analysis of API and ANGA Survey Responses 16

23. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED Also, the ANGA/API survey collected information on the number of workovers forvertical and horizontal unconventional gas wells. Nearly 99% of the unconventional gas wellworkovers were on vertical wells. Additionally, 18% of the gas well workovers from theAPI/ANGA survey were conducted on gas wells without hydraulic fracturing. A second phase of the survey was conducted which targeted collecting gas well re-fracture information for 2010 to provide a better estimate than EPAs assumption that 10% ofwells are re-fractured each year. This portion of the ANGA/API survey requested informationjust for “unconventional” gas wells (i.e., those located on shale, coal-bed methane, and tightformation reservoirs), where the formations require fracture stimulation to economically producegas. A re-fracture or workover was defined for this second phase of the survey as a re-completion to a different zone in an existing well or a re-stimulation of the same zone in anexisting well. These results are summarized in Table 8. While there likely is significant overlap of unconventional well data reported in the firstand second phases of the survey (which covered over 62,500 unconventional wells and 69,000unconventional wells respectively), combined these data indicate an unconventional well re-fracture rate of 1.6% to 2.3% including AAPG 540 and 0.7% to 1.15% excluding AAPG 540. AAPG Basin 540 (i.e. DJ Basin) which is part of the Rocky Mountain Region stands outin Tables 7 and 8. After four (4) to eight (8) years of normal production decline, the gas wells inthis basin can be re-fractured in the same formation and returned to near original production.Success of the re-fracture program in the DJ Basin is uniquely related to the geology of theformation, fracture reorientation, fracture extension and the ability to increase fracturecomplexity. Also, most DJ Basin gas wells are vertical or directional, which facilitates theability to execute re-fracture operations successfully and economically. These characteristicsresult in a high re-fracture or workover rate specific to this formation. ANGA and API believe the high re-fracture rate observed in the DJ Basin is unique andnot replicated in other parts of the country. There may be a few other formations in the worldthat have similar performance, but the successful re-fracture rate in the DJ Basin is not going tobe applicable to every asset/formation and there is no evidence of the high re-fracture rate in anyof the other 22 AAPGs covered in the API/ANGA survey. It is highly dependent on the type ofrock, depositional systems, permeability, etc. For these reasons, re-fracture rates for tight gaswells and all gas wells with and without AAPG Basin 540 are summarized in Tables 7 and 8.Summary and Analysis of API and ANGA Survey Responses 17

24. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED TABLE 8. API/ANGA SURVEY – SUMMARY OF 2010 GAS WELL WORKOVERS ON UNCONVENTIONAL WELLS BY AAPG BASIN AND NEMS REGION (SECOND PHASE SURVEY DATA) Number of Hydraulic Fracture Number of Workovers on Regional % Unconventional Previously % Wells re- Wells re- NEMS Operating Gas Fracture fractured fractured Region AAPG Wells Stimulated Wells per year per year 160 1,976 0 0.00% Northeast 0% 160A 760 0 0.00% 200 2 0 0.00% 220 649 2 0.31% 222 629 3 0.48% Gulf Coast 0.91% 230 820 4 0.49% 250 13 0 0.00% 260 2,830 36 1.27% 345 3,296 11 0.33% Mid- 350 213 3 1.41% Continent 355 282 8 2.84% 0.95% 360 7,870 89 1.13% 375 12 0 0.00% 385 1 0 0.00% 400 64 0 0.00% 415 1,834 0 0.00% 420 838 8 0.95% Southwest 1.04% 430 1,548 36 2.33% 435 2 0 0.00% 515 1 0 0.00% Rocky 540 5,950 866 14.55% 4.7% Mountain 580 8,197 8 0.10% 595 5,222 32 0.61% Not specified 26,025 487 1.87% 1.87% Unconventional TOTAL (all wells) 69,034 1,593 2.31% Unconventional Median 790 3 Rocky Mountain Region Unconventional Total 19,370 906 4.68% Unconventional TOTAL (Without AAPG 540) 63,084 727 1.15%Summary and Analysis of API and ANGA Survey Responses 18

25. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED 4.2 WRAP Survey Other information on re-fracture rates is available in a survey conducted by the WesternRegional Air Partnership (WRAP). WRAP conducted a survey of production operators in theRocky Mountain Region (Henderer, 2011) as part of the initiative to develop GHG reportingguidelines for a regional GHG cap and trade program. Within each basin in this region, the top oil and gas producers were identified and invitedto participate in the survey. The goal was to have operator participation that represented 80% ofthe production for the region. The spreadsheet survey requested information on the completions,workovers, and emissions associated with these activities. An emission factor and frequency ofre-fracturing was developed for each basin as a weighted average of the operator responses. The re-fracture rates from the WRAP survey are shown in Table 9 (Henderer, 2011).TABLE 9. WRAP SURVEY – SUMMARY OF GAS WELL WORKOVERS BY AAPG BASIN FOR THE ROCKY M OUNTAIN REGION, 2006 DATA # Wells represented # Wells % AAPG Basin by survey Recompleted Recompleted 515 4,484 121 2.70% 530 731 5 0.68% 535 4,982 201 4.03% 540 8,247 636 7.71% 580 3,475 14 0.40% 595 4,733 275 5.81% Total 26,652 1,252 Weighted average 4.70% AAPG Basin 540 results in the highest re-fracture rate for this data set, consistent withthe ANGA/API survey as noted above. It is noteworthy that, while there are differences amongindividual AAPG Basin results, the weighted average re-fracture rate from the WRAP survey in2006 is the same as the Rocky Mountain regional 4.7% re-fracture rate from the API/ANGAsurvey shown in Table 8. 4.3 Impact of Completions and Re-fracture Rate Assumptions Table 10 compares the considerable reduction in the national GHG inventory that wouldresult from applying a lower re-fracture rate. EPA indicated that the national inventory assumes 10% of unconventional gas wells arere-fractured each year. Table 10 replaces this value with results from the ANGA/API survey. Are-fracture rate of 1.15% is applied to unconventional gas wells in the Mid-Continent andSouthwest regions (No unconventional gas wells were assigned to the Northeast and Gulf Coastregions. The West Coast region is not shown since the API/ANGA survey did not include anyresponses for gas well operations in this region.) A re-fracture rate of 4.7% is applied tounconventional gas wells in the Rocky Mountain region.Summary and Analysis of API and ANGA Survey Responses 19

26. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED With these adjustments to the re-fracture rate for unconventional gas wells, thenational emission estimate is reduced by 72% for this emission source category, from 712,605metric tons of CH 4 to 197,311 metric tons of CH 4 when compared on a consistent basis. 4.4 Completion and Re-fracture Emission Factor In the 2009 GHG national inventory, EPA applies an emission factor of 2,454 scfCH 4 /event for conventional gas well workovers, while the emission factor for unconventionalgas well completions and workovers was increased to 7,623,000 scf CH 4 /event (EPA, 2011b).Similarly, for the 2010 national GHG inventory, EPA maintained the emission factor of 2,454scf CH 4 /event for gas well workovers without hydraulic fracturing, but applied an averageemission factor of 7,372,914 to gas well workovers with hydraulic fracturing (EPA, 2012).(EPA applies slightly different emission factors for each NEMS region based on differing gascompositions.) The ANGA/API survey focused on activity data and did not collect data to revise theemission factor for unconventional gas well completions and workovers.Summary and Analysis of API and ANGA Survey Responses 20

27. TABLE 10. API/ANGA SURVEY –GAS WELL WORKOVER EMISSIONS COMPARISON Revised Emissions, 2010 EPA National tonnes 2010 EPA Adjusted # Inventory CH 4 National workovers Estimated API & ANGA - EPA (based on EPA Inventory (based on Emission Emissions, ANGA/API # API/ANGA Factor, scf tonnes survey)NEMS Region Well type workover survey) CH 4 /workover CH 4 * % Difference Wells without HydraulicNortheast 8,208 8,208 2,607 409 409 Fracturing Wells with Hydraulic 0 0 7,694,435 0 0 Fracturing Wells without HydraulicMid Continent 3,888 3,888 2,574 191 191 Fracturing Wells with Hydraulic 1,328 153 7,672,247 194,950 22,462** -89% Fracturing Wells without Hydraulic 3,822 3,822 2,373 174 174 FracturingRocky Wells with Hydraulic 2,342 1,100 7,194,624 322,402 151,432** -53%Mountain Fracturing Wells without HydraulicSouthwest 1,803 1,803 2,508 87 87 Fracturing Wells with Hydraulic 1,374 158 7,387,499 194,217 22,382** -89% Fracturing Wells without HydraulicGulf Coast 3,300 3,300 2,755 174 174 Fracturing Wells with Hydraulic 0 0 8,127,942 0 0 FracturingTOTAL 712,605 197,311 -72% * EPA Estimated emissions = 2010 # Workovers x EPA 2010 Emission Factor, converted to mass emissions based on 60°F and 14.7 psia. ** Revised emissions = Adjusted # Workovers x Emission Factor, converted to mass emissions based on 60°F and 14.7 psia.Summary and Analysis of API and ANGA Survey Responses 21

28. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDEmissions Data from WRAP Study The WRAP study discussed in Section 4.2 also gathered data on emissions fromcompletions. This information supports a revised emission factor but was reported by sourcesoutside the ANGA/API data survey. The results are summarized in Table 11. The WRAPemission factor is 78% lower than EPA’s emission factor (9.175 MMscf gas/event). The WRAPsurvey did not provide a methodology for determining emissions data. TABLE 11. WRAP SURVEY – SUMMARY OF COMPLETION EMISSIONS FOR THE ROCKY MOUNTAIN REGION, 2006 DATA Weighted average gas emissions from # completion, Mcf completions AAPG Basin gas/well represented 515 167 207 530 268 54 535 76 642 540 59 608 580 6,559 283 595 4,053 819 Total 2,613 Weighted average 2,032 Mcf/well 4.5 Data Limitations for Completion and Re-fracture Emissions Although the data sets are limited, it appears that EPA’s assumed re-fracture rate of 10%is a significant overestimate. Information from the API/ANGA survey indicates that evenincluding what appears to be unique activity in AAPG-540, the re-fracture rate is much lessfrequent, ranging from 1.6% to 2.3% based on two sets of survey information (Tables 7 and 8,respectively). The re-fracture rate for AAPG Basin 540 appears to be higher than other areas inthe U.S. due to unique geologic characteristics in that region (4.7% based on a weighted averageof data reported for that region). Without AAPG Basin 540, the national rate of re-fracturing isbetween 0.7% and 1.15% of all gas wells annually. Additionally, limited information on the emissions from completions and workovers withhydraulic fracturing indicate that EPA’s GHG emission factor for these activities is significantlyoverestimated. It is expected that better emissions data will develop as companies begin tocollect information for EPA’s mandatory GHG reporting program (EPA, 2011c).Summary and Analysis of API and ANGA Survey Responses 22

29. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED 5. Other Surveyed Information EPA had indicated that activity data for centrifugal compressor wet seals and pneumaticdevices used in the national inventory is lacking. Note that the need for better equipment datapersists throughout the majority of the U.S. inventory and is not unique to the oil and natural gasindustry. The ANGA/API survey requested the following information related to centrifugalcompressors and pneumatic devices: • The number of centrifugal compressors, reported separately for production/gathering versus processing; • The number of centrifugal compressors with wet versus dry seals, reported separately for production/gathering versus processing; • The number of pneumatic controllers, classified as “high-bleed,” “low-bleed,” and “intermittent,” reported separately for well sites, gathering/compressor sites, and gas processing plants; and • The corresponding number of well sites, gathering/compressor sites, and gas processing plants, associated with the pneumatic controller count. 5.1 Centrifugal CompressorsProcessing Facilities The API/ANGA survey collected the equivalent of 5% of the national centrifugalcompressor count for gas processing operations (38 centrifugal compressors from the survey,compared to 811 from EPA’s 2010 national GHG inventory). For the gas processing centrifugalcompressors reported through the survey, 79% were dry seal compressors and 21% were wetseals. EPA’s 2010 national inventory reported 20% of centrifugal compressors at gas processingplants were dry seal, and 80% were wet seal. EPA’s emission factor for wet seals (51,370 scfdCH 4 /compressor) is higher than the emission factor for dry seals (25,189 scfd CH 4 /compressor). 6 Based on the ANGA/API survey, EPA appears to be overestimating emissions fromcentrifugal compressors. If the small sample size from the API/ANGA survey is representative,non-combustion emissions from centrifugal compressors would be 173,887 metric tons ofmethane compared to 261,334 metric tons of methane from the 2010 national inventory (whenapplying industry standard conditions of 60 °F and 14.7 psia to convert volumetric emissions tomass emissions). Although based on very limited data, if the ANGA/API survey results reflectthe population of wet seal versus dry seal centrifugal compressors, the emissions from thissource would be reduced by 34% from EPA’s emission estimate in the national inventory. Betterdata on the number of centrifugal compressors and seal types will be available from companiesreporting to EPA under the mandatory GHG reporting program.6 EPA Table A-123, of Annex 3 of the 2010 inventory report.Summary and Analysis of API and ANGA Survey Responses 23

30. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDProduction and Gathering Facilities Very few of the data sets reported through the API/ANGA survey indicate counts ofcentrifugal compressors associated with production/gathering operations - only 550 centrifugalcompressors from 21 participating companies. EPA’s 2010 GHG inventory did not includecentrifugal compressors in production/gathering operations. On a well basis, the surveyresponses equate to 0.07 centrifugal compressors per gas well, with 81% dry seal centrifugalcompressors and the remaining wet seal compressors. Information reported through EPA’smandatory GHG reporting program will provide additional information to account for GHGemissions from centrifugal compressors in production operations. 5.2 Pneumatic Controllers Table 12 summarizes the survey responses for pneumatic controllers. For each type oflocation – gas well sites, gathering compressor sites, and gas processing plants – the count of thenumber of sites represented by the survey data is shown. Table 12 also shows the percent ofeach pneumatic controller type for each type of location. TABLE 12. ANGA/API SURVEY –PNEUMATIC CONTROLLER COUNTS Gathering/ Gas Compressor Processing Gas Well Sites Sites Plants # wells, sites or plants 48,046 wells 1,988 sites 21 plants # controllers/well, site or 0.99 per well 8.6 per site 7.8 per plant plant # Low Bleed Controllers 12,850 27% 5,596 33% 117 71% # High Bleed Controllers 11,188 24% 1,183 7% 47 29% # Intermittent Controllers 23,501 49% 10,368 60% 0 0% The survey requested that the responses designate pneumatic controllers as either “highbleed”, “low bleed”, or “intermittent” following the approach each company is using for SubpartW reporting. For example, Subpart W defines high-bleed pneumatic devices as automated,continuous bleed flow control devices powered by pressurized natural gas where part of the gaspower stream that is regulated by the process condition flows to a valve actuator controller whereit vents continuously (bleeds) to the atmosphere at a rate in excess of 6 standard cubic feet perhour (EPA, 2011c). EPA does not currently track pneumatic controllers by controller type in the nationalinventory. This information will be collected under 40 CFR 98 Subpart W starting in September2012. From the API/ANGA survey, intermittent bleed controllers are the more prevalent type atgas well sites and gathering/compressor sites, while gas plants predominately use low-bleedcontrollers. No intermittent controllers were reported for gas plants by the survey respondents. Table 13 compares emission results based on applying the emission factors from theEPA’s GHG reporting rule to emissions presented in the 2010 national GHG inventory, using thecounts of pneumatic controller from the ANGA/API survey for production operations.Summary and Analysis of API and ANGA Survey Responses 24

31. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED For production, the EPA national inventory combines pneumatic controller countsassociated with large compressor stations with pneumatic controllers in production. An emissionfactor for each NEMS region is applied to the count of total controllers in each NEMS region.For this comparison, a weighted average emission factor of 359 scfd CH 4 /device was applied tothe count of pneumatic controllers located at well sites and gathering/compressor sites. Under the EPA mandatory reporting rule (40 CFR 98 Subpart W), separate emissionfactors are applied to pneumatic controllers based on the controller type and whether thecontroller is located in the Eastern or Western region of the United States, as specified in the rule(EPA, 2011c). For this comparison, an average of the eastern and western emission factors isapplied to each device type in computing the emission estimates resulting from the EPA GHGreporting rule. TABLE 13. PNEUMATIC CONTROLLER EMISSION COMPARISON – PRODUCTION OPERATIONS API/ANGA Survey EPA GHG Reporting Rule 2010 National GHG Count of Controllers (Subpart W) Inventory Emission Emission Gas Gathering/ Factor,* Emissions, Factor, Emissions, Well Compressor scfh tonnes scfd tonnes Sites Sites Total CH 4 /device CH 4 /yr CH 4 /device CH 4 /yr # Low Bleed 12,850 5,596 18,446 1.58 4,885 46,286 Controllers # High Bleed 11,188 1,183 12,371 42.35 87,814 359 31,042 Controllers # Intermittent 23,501 10,368 33,869 15.3 86,856 84,987 Controllers Total 64,686 179,556 162,315* Emission factors shown are the average of the eastern and western emission factors from Table W- 1A (EPA, 2011c). Based on the types of pneumatic controllers reported in the ANGA/API survey, EPA’smandatory GHG reporting rule could increase CH 4 emissions 11% over the pneumatic controllerportion of the 2010 national GHG inventory. To put this in context, in EPA’s inventory reportfor 2010, emissions from pneumatic controllers accounted for approximately 13% of CH 4emissions from the natural gas field production stage. Any increase from that initially reporteddata, however, will likely represent a worst case scenario. It is important to remember thatpneumatic controllers operate only intermittently, so variability such as the frequency andduration of the activations will be important information to consider when defining an accurateand effective reporting regime for these sources. EPA’s mandatory GHG reporting rule does not require reporting emissions frompneumatic controllers at gas processing plants, so no emission factors are specified. The GHGnational inventory applies an emission factor of 164,721 scfy CH 4 per gas plant for pneumaticcontrollers. For the national inventory, this results in 1,856 tonnes CH 4 emissions - a very smallcontribution to CH 4 emissions from onshore oil and gas operations.Summary and Analysis of API and ANGA Survey Responses 25

32. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED 6. Conclusions API and ANGA members believe this to be the most comprehensive set of natural gasdata to date and are pleased to share these results with both regulators and the public. Based on the information gathered from member companies during this project, itappears that EPA has overstated several aspects of GHG emissions from unconventional naturalgas production. As summarized in Table 14, the ANGA/API survey data results in significantlylower emission estimates for liquids unloading and unconventional gas well refracturing whencompared to EPA’s emission estimates in the national inventory. Using the combined emissionestimates from the survey for these two key emission sources would indicate a 50% reduction incalculated natural gas production sector emissions compared to EPA’s estimates. This reductionwould shift Natural Gas Systems from the largest to the second largest producer of methaneemissions (approximately 123.4 MMT CO 2 e in lieu of 215.4 MMT CO 2 e), behind EntericFermentation (which is a consequence of bovine digestion, at 141.3 MMT CO 2 e). TABLE 14. EMISSION COMPARISON BETWEEN EPA AND INDUSTRY DATA Impact on Source Source Category Category EPA National Inventory API/ANGA Survey Emissions API & ANGA - EPA EPA % of % of EPA Revised Metric tons of CH4 Production Metric tons of CH4 Production % Difference in Total Total Emissions Liquids Unloading 4,501,465 * 51% 637,766 14% -86% Unconventional Well Re-fracture 712,605 * 8% 197,311 4% -72% Rates Other Production ** 3,585,600 41% 3,585,600 81% Sector Emissions Total Production 8,799,670 4,420,677 -50% Sector Emissions * EPA’s estimates are adjusted to industry standard conditions of 60 degrees F and 14.7 psia for comparison to the ANGA/API emission estimates. ** The “Other Production Sector Emissions” are comprised of over 30 different source categories detailed in Table A-129 in the Annex of the EPA’s 2012 national inventory. The “Other Production Sector Emissions” are the same values for this comparison between the EPA national inventory and the API/ANGA survey to focus the comparison on quantified differences in emission estimates for gas well liquids unloading and unconventional well re-fracture rates. This project was directed toward gathering more robust information on workovers,completions, liquids unloading, centrifugal compressors, and pneumatic controllers with theintent of supporting revisions to the activity factors used in EPA’s national inventory and citedSummary and Analysis of API and ANGA Survey Responses 26

33. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDby many media publications. Although limited information was collected on centrifugalcompressors and pneumatic controllers, the survey results indicated potential additionaldifferences, which are not included in the Table 14 comparison, when comparing total emissionsfrom all sources to the national inventory. Additional future data collection efforts, includingmore detailed reporting under Subpart W of the GHGRP will likely resolve these differences andcontinue to inform the overall natural gas emissions data. In the meantime, however, while API and ANGA recognize that the data collected forthis report represents a sample of the universe of natural gas wells operating in the U.S., webelieve that the conclusions drawn from the data analysis are relevant and representative ofnatural gas production as whole. In EPA’s gas well count, 21 of the AAPG basins each havemore that 1% of the total well count. The ANGA/API survey has wells from 19 of those 21basins. In terms of wells represented by these basins, 92% of the total EPA well count isaccounted for by wells in those 21 basins, while 95% of the API/ANGA surveyed gas wells areaccounted for by those 21 basins. This indicates that the ANGA/API survey results have goodrepresentation for the basins with the largest numbers of wells nationally. Industry also believes that the systematic approach in which the API/ANGA data werecollected and vetted by natural gas experts is an improvement over the ad hoc way in which EPAcollected some of their data. This study indicates that EPA should reconsider their inventorymethodologies for unconventional natural gas production particularly in light of morecomprehensive and emerging data from the industry. ANGA and API members look forward toworking with the agency to continue to educate and evaluate the latest data as it develops aboutthe new and fast-changing area of unconventional well operations.Summary and Analysis of API and ANGA Survey Responses 27

34. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED 7. ReferencesEnergy Information Administration (EIA). "Number of Producing Gas Wells”, U.S. and Statelevel data, annual, 2010 Data, Released February 29, 2012.http://www.eia.gov/dnav/ng/ng_prod_wells_s1_a.htmHenderer, Douglas. Personal communications on the survey instrument utilized for the WRAPIII inventory. KleinFelder, Littleton, Colorado, dhenderer@kleinfelder.com, October, 2011.http://www.wrapair.org/forums/ogwg/PhaseIII_Inventory.htmlIHS CERA, “Mismeasuring Methane,” 2011. http://www.ihs.com/info/en/a/mis-measuring-methane-report.aspxU.S. Environmental Protection Agency, Natural Gas STAR Lessons Learned: Installing PlungerLift Systems In Gas Wells, 2006. http://www.epa.gov/gasstar/documents/ll_plungerlift.pdfUnited States Environmental Protection Agency (EPA). “Greenhouse Gas Emissions Reportingfrom the Petroleum and Natural Gas Industry, Background Technical Support Document,” U.S.Environmental Protection Agency, Climate Change Division, Washington DC, November, 2010.United States Environmental Protection Agency (EPA). “EPA U.S. Oil and Gas GHGInventory, July 14, 2011 Webcast, Review Inventory Method, Potential Inventory Improvementon Well Completions and Workovers,” Distributed for discussion purposes only. Final Draft,July 15, 2011(a).United States Environmental Protection Agency (EPA). Inventory of Greenhouse Gas Emissionsand Sinks: 1990-2009, Washington DC, April, 2011(b)http://www.epa.gov/climatechange/emissions/downloads11/US-GHG-Inventory-2011-Complete_Report.pdfUnited States Environmental Protection Agency (EPA). Inventory of Greenhouse Gas Emissionsand Sinks: 1990-2010, Washington DC, April, 2012.http://www.epa.gov/climatechange/emissions/usinventoryreport.htmlUnited States Environmental Protection Agency (EPA). “Mandatory Reporting of GreenhouseGases: Technical Revisions to the Petroleum and Natural Gas Systems Category of theGreenhouse Gas Reporting Rule”, Final Rule, Federal Register, Vol. 76, No. 247, December 23,2011(c).Summary and Analysis of API and ANGA Survey Responses 28

35. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDUnited States Environmental Protection Agency (EPA). “Supplement to Appendix D of theRevisions to Subpart I and Subpart W Technical Support Document – Listing of Well Count byGroup Type 2010: Well Counts by Group.pdf”. Supporting information provided by EPA withthe pre-Federal Register version of amendments to Subpart W. August 22, 2011(d).http://epa.gov/climatechange/emissions/downloads11/documents/Well-Counts-by-Group.pdfUnited States Environmental Protection Agency, Technical Note on the 1990 to 2009 InventoryEstimates for Natural Gas Systems, Washington DC, 2011(e).http://www.epa.gov/outreach/downloads/TechNote_Natural%20gas_4-15-11.pdfZeller, Tom Jr. “Studies Say Natural Gas Has Its Own Environmental Problems,” New YorkTimes, April 11, 2011.Summary and Analysis of API and ANGA Survey Responses 29

36. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDAppendix A. API/ANGA Survey FormsThe following provides the survey forms used to gather data presented in this report. FIGURE A-1. SURVEY INSTRUCTIONSSummary and Analysis of API and ANGA Survey Responses 30

37. FIGURE A-2. GAS WELL SURVEY DATASummary and Analysis of API and ANGA Survey Responses 31

38. FIGURE A-3. GAS WELL WORKOVER SURVEY DATASummary and Analysis of API and ANGA Survey Responses 32

39. FIGURE A-4. GAS WELL LIQUIDS UNLOADING SURVEY DATASummary and Analysis of API and ANGA Survey Responses 33

40. FIGURE A-5. OTHER S URVEY DATASummary and Analysis of API and ANGA Survey Responses 34

41. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDAppendix B. ANGA/API Well Survey InformationResponses from the API/ANGA survey covered more than 60,000 wells and provided data on: • # of gas wells without hydraulic fracturing (anytime in their history) • # of gas wells with hydraulic fracturing (any time in their history); ◦ # of vertical gas wells with hydraulic fracturing (anytime in their history); ◦ # of horizontal gas wells with hydraulic fracturing (anytime in their history); • # of completions for vertical gas wells with hydraulic fracturing; • # of completions for horizontal gas wells with hydraulic fracturing; • # of completions for gas wells without hydraulic fracturing; • # of workovers for vertical wells with hydraulic fracturing; • # of workovers for horizontal wells with hydraulic fracturing; and • # of workovers for wells without hydraulic fracturing. Table B-1 summarizes the well data collected by the ANGA/API survey and presents itsdistribution by formation type and region. The regional distribution follows the National EnergyModeling System (NEMS) regions defined by the EIA. The data are compared to EPA’snational well counts classified by type as provided in the August 2011 database file (EPA,2011d).Summary and Analysis of API and ANGA Survey Responses 35

42. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED TABLE B-1. API/ANGA SURVEY – SUMMARY OF GAS WELL COUNTS BY TYPE AND NEMS REGION* Conventional Coal-bed NEMS Region Wells Shale Methane Tight Unspecified Northeast 12,144 3,541 9 3,874 2,563 Gulf Coast 2,870 1,990 - 7,968 1,521 Mid-Continent 9,081 2,333 - 3,747 5,579 Southwest 646 1,208 - 726 2,326 Rocky Mountain 3,707 366 5,458 18,053 11 West Coast - - - - - Unspecified 1,307 Survey TOTAL 28,448 9,438 5,467 34,368 13,307 % of EPA 2010 Well Counts (from database file) 14.2% 30.1% 11.5% 45.6% Overall Survey Total 91,028 200,921 31,381 47,371 75,409 EPA Well Counts (2010, from 56.6% 8.8% 13.3% 21.2% database file) 355,082 EPA National Inventory (2010) 484,795 EIA National Well Count (2010) 487,627 * ANGA/API survey data represents well counts current for calendar year 2010 or the first half of 2011. As shown in Table B-1, data from the API/ANGA survey represent approximately 26%of the national gas wells reported by EPA’s database (or 18.7% of the EIA well count data).This includes almost 46% of all tight gas wells and 30% of shale gas wells. This may indicatethat the ANGA/API information has an uneven representation of unconventional gas wells, andin particular shale and tight gas wells, but it also appears that EPA’s data may mis-categorizethese types of wells. For example, the EPA/HPDI data set contains few wells from Pennsylvaniaand West Virginia while the API/ANGA survey includes 9,422 wells from that area (AAPG160A).Summary and Analysis of API and ANGA Survey Responses 36

43. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDTable B-2 summarizes additional details on the natural gas wells information collected throughthe second data collection effort by the ANGA/API survey which covered 60,710 wells. TABLE B-2. ANGA/API SURVEY – ADDITIONAL DETAILS ON GAS WELL COUNTS* # Wells w/out hydraulic # Wells with hydraulic fracturing fracturing (any time in their history) (anytime in # Horizontal their history) Total # Vertical wells wells TOTAL Conventional 1,498 16,678 14,844 1,834 TOTAL Coal Bed Methane 42 3,475 3,424 42 TOTAL Shale 1,931 9,084 2,012 7,072 TOTAL Tight 122 27,880 24,048 3,835 TOTAL OVERALL 3,593 57,117 44,325 12,783 * API/ANGA survey data represents well counts current for calendar year 2010 or the first half of 2011. Additional information on natural gas wells with and without hydraulic fracturing wasprovided for approximately two-thirds (60,710 natural gas wells) of the total well data collectedby the ANGA/API survey. For this subset of the well data, 94% of the gas wells have beenhydraulically fractured at some point in their operating history, including almost 92% of theconventional wells. EPA’s 2010 national inventory reported 50,434 gas wells with hydraulicfracturing. This is very similar to the number of unconventional gas wells that EPA reported inthe 2009 national inventory. Based on the API/ANGA survey results, it appears that EPA hasunderestimated the number of gas wells with hydraulic fracturing. Of the ANGA/API survey responses for wells that have been hydraulically fractured,most (77.6%) are vertical wells. Vertical wells are predominately conventional gas wells, coal-bed methane and tight gas wells; while the majority of shale gas wells are horizontal. EPA doesnot currently distinguish between vertical and horizontal gas wells.A Short Note About EPA and EIA’s Well Counts There is a discrepancy of over 132,000 natural gas wells between the EPA databaseinformation (EPA, 2011d) and the EIA national gas well counts (EIA, 2012), and a difference ofalmost 130,000 gas wells between the two EPA data sources (EPA, 2011d and EPA, 2012). Thisdifference needs to be understood since ultimately both the IHS (EIA) and HPDI (EPA) dataoriginate from the same state-level sources of information. The EIA provides a gas well count of 487,627 for 2010 based on Form EIA-895A 7, theBureau of Ocean Energy Management, Regulation and Enforcement (formerly the Minerals7 Form EIA-895, Annual Quantity And Value Of Natural Gas Production Report;http://www.eia.gov/survey/form/eia_895/form.pdfSummary and Analysis of API and ANGA Survey Responses 37

44. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDManagement Service) data, and World Oil Magazine (EIA, 2010). However, the EIA does notclassify gas wells by conventional and unconventional, or by formation types, precluding moredetailed comparison against the EIA data. For some parameters the classifications were based onqualitative descriptions of the formations’ physical properties (e.g. permeability) rather than onactual measurements (i.e. permeability data in millidarcy readings). 8 EPA provides a similar well count in the 2010 national inventory: 434,361 non-associated gas wells + 50,434 gas wells with hydraulic fracturing, resulting in a total of 484,795gas wells (EPA, 2012). Further classification of gas wells or description on what constitutes a“non-associated” gas well versus a “gas well with hydraulic fracturing” is not provided in EPA’snational inventory. Small differences in the HPDI and IHS original data may arise from definitionaldifferences as HPDI and IHS compile the raw data. In addition, each state may have a differentinterpretation of well definitions of gas versus oil wells that introduces differences among statesfor the wells reported. EPA had indicated in discussions with the API/ANGA group that theirdatabase well count information may not include all of the wells in the Marcellus basin. EIAindicates 44,500 gas wells in Pennsylvania in 2010. However, even in accounting for thesewells, there is still a large difference (almost 88,000 wells) between EPA’s total gas well numberfrom their database source and EIA’s well data. Nevertheless, these discrepancies among the well counts need to be understood sincethese data all originate from the same state-level sources of information. Differences could arise,for example, from different interpretations of well definitions. Since the EIA data is the de facto benchmark in the energy industry, the differencebetween the EIA and EPA well count data needs to be understood before any meaningfulconclusions can be made from the EPA data. Since EPA’s well count from HPDI was much lower than the EIA, this report does notattempt to come up with a national gas well count but chose to use the 355,082 number from theEPA HPDI database because it was the only available database which parsed the wells intoconventional and unconventional categories (EPA, 2011d).8 Information provided by Don Robinson of ICF (EPA’s contractor).Summary and Analysis of API and ANGA Survey Responses 38

45. Appendix C. Emission Estimates for Gas Well Liquids UnloadingTables C-1 through C-4 summarize the liquids unloading emissions data collected through the API/ANGA survey and the resultingemission estimates. The emission factors reported in Table 4 are based on a regional weighted average of the conventional andunconventional gas wells, with and without plunger lifts. This provided a consistent comparison against the EPA emission factorswhich are reported only on a regional basis and do not differentiate between conventional and unconventional wells or wells with andwithout plunger lifts. TABLE C-1. LIQUIDS UNLOADING FOR CONVENTIONAL GAS WELLS WITHOUT PLUNGER LIFTSNEMS Region Northeast Gulf Coast Mid-Continent Southwest# venting gas wells 190 916 12 6 1 38 220# gas well vents 4,335 39,668 144 60 1 2,444 880Average casing diameter, inches 5 4.5 5.5 3.65 4.83 4 5.5Average well depth, feet 3,375 3,448 10,000 19,334 7,033 4,269 8,000Average surface pressure, psig 85 50 Applied 224 25.5 60.8 100(for venting wells) average 122Average venting time, hours 1 2 1 2.5 .25 4.95 1Average gas flow rate, Mscfd 2,861 7,388.5 300 664 58.43 84 100Total emissions, scf gas/yr 11,503,329 51,547,287 1,961,463 1,322,380 1,548 3,769,194 7,879,520Emissions per well, scfy gas/well 60,544 56,274 163,455 220,397 1,548 99,189 35,816Summary and Analysis of API and ANGA Survey Responses 39

46. TABLE C-2. LIQUIDS UNLOADING FOR CONVENTIONAL GAS WELLS WITH PLUNGER LIFTS NEMS Region Northeast Mid-Continent # venting gas wells 33 109 164 2 10 # gas well vents 1,272 4,217 489,912 23 7,300 Average tubing diameter, inches 2 2.375 1.995 2 2.375 Average well depth, feet 3,375 3,448 4,269 7,033 9,500 Average surface pressure, psig (for 85 50 60.8 25.5 500 venting wells) Average venting time, hours 1 0.3 0.067 0.75 0.08 Average gas flow rate, Mscfd 2,861 7,388.5 84 58.43 30 Total emissions, scf gas/yr 599,664 1,517,294 187,255,825 6,713 72,367,809 Emissions per well, scfy gas/well 18,172 13,920 1,141,804 3,357 7,236,781Summary and Analysis of API and ANGA Survey Responses 40

47. TABLE C-3. LIQUIDS UNLOADING FOR UNCONVENTIONAL GAS WELLS WITHOUT PLUNGER LIFTS NEMS Region Northeast Gulf Coast # venting gas wells 337 6 14 8 27 11 15 # gas well vents 27,720 6 14 104 207 572 15 Average casing diameter, 4.5 5.5 5.5 5.5 4.5 5.5 10.75 inches Average well depth, feet 4,845 6,000 8,500 11,000 9,000 13,752 16,000 Average surface pressure, psig 121.6 400 3,200 200 50 450 1,671 (for venting wells) Average venting time, hours 1.3638 3 4 1 5.3 2 2 Average gas flow rate, Mscfd 26 200 13,000 25 130 353 8,500 Total emissions, scf gas/yr 122,362,610 177,839 5,887,104 2,560,844 722,663 39,633,526 17,501,885 Emissions per well, scfy 363,094 29,640 420,507 320,106 26,765 3,603,048 1,166,792 gas/wellSummary and Analysis of API and ANGA Survey Responses 41

48. TABLE C-3. LIQUIDS UNLOADING FOR UNCONVENTIONAL GAS WELLS WITHOUT PLUNGER LIFTS, CONTINUED NEMS Region Gulf Coast Mid-Continent # venting gas wells 146 2 10 40 177 3 136 215 # gas well vents 146 12 120 40 400 7.2 391.2 2,580 Average casing diameter, 4.5 5.5 5.5 8.625 5.5 4.92 5.02 5.5 inches Average well depth, feet 8,500 11,647 11,000 12,500 3,911 10,293 7,888 11,000 Average surface pressure, 15 25 94 661 80 90.04 98.75 200 psig (for venting wells) Average venting time, hours 0.6875 1.5 4 1 2.5 1.58 1.925 0.5 Average gas flow rate, Mscfd 99 83 92 6,500 250 727 875 100 Total emissions, scf gas/yr 139,473 40,837 1,400,265 9,096,858 1,416,389 77,333 2,874,991 63,528,630 Emissions per well, scfy 955 20,418 140,027 227,421 8,002 25,778 21,140 295,482 gas/wellSummary and Analysis of API and ANGA Survey Responses 42

49. TABLE C-3. LIQUIDS UNLOADING FOR UNCONVENTIONAL GAS WELLS WITHOUT PLUNGER LIFTS, CONTINUED NEMS Region Southwest Rocky Mountain # venting gas wells 228 6 3 113 2 28 # gas well vents 221 6 1 2,004 4 10,584 Average casing diameter, 9.625 5.5 5 4.038 4.7 4.5 inches Average well depth, feet 8,725 8,000 15,000 11,149 11,056 10,844 Average surface pressure, psig 208 50 200 250 250 198 (for venting wells) Average venting time, hours 1 0.5 6.67 1.616 0.75 3.18 Average gas flow rate, Mscfd 1,500 12 150 127 433 83 Total emissions, scf gas/yr 13,747,516 26,862 63,188 33,701,560 90,364 170,274,852 Emissions per well, scfy 60,296 4,477 21,063 298,244 45,182 6,081,245 gas/wellSummary and Analysis of API and ANGA Survey Responses 43

50. TABLE C-4. LIQUIDS UNLOADING FOR UNCONVENTIONAL GAS WELLS WITH PLUNGER LIFTS NEMS Region Northeast Gulf Coast # venting gas wells 308 103 5 3 2 22 59 5 # gas well vents 63,840 75,190 194 156 2 22 354 5 Average tubing diameter, 2.375 2.375 2.375 2.375 2.375 2.375 2.375 2.375 inches Average well depth, feet 4,845 2,500 7,000 13,752 16,000 8,500 11,647 12,500 Average surface pressure, 121.6 200 130 450 1,671 15 25 661 psig (for venting wells) Average venting time, 0.2209 0.05 0.1 2 1 0.875 0.3 0.5 hours Average gas flow rate, 26 15 628 353 8,500 99 83 6,500 Mscfd Total emissions, scf gas/yr 78,496,300 78,461,940 368,444 2,036,862 288,681 7,401 215,123 86,220 Emissions per well, scfy 254,858 761,766 73,689 678,954 144,341 336 3,646 17,244 gas/wellSummary and Analysis of API and ANGA Survey Responses 44

51. TABLE C-4. LIQUIDS UNLOADING FOR UNCONVENTIONAL GAS WELLS WITH PLUNGER LIFTS, CONTINUED NEMS Region Mid-Continent Southwest # venting gas wells 48 4 64 29 18 # gas well vents 155,742 9.6 170.4 348 25 Average tubing diameter, inches 2.375 3.88 4.11 2.4 1.995 Average well depth, feet 3,911 10,293 7,888 Applied average 8,725 9,521 Average surface pressure, psig (for 80 90.04 98.75 74.69 208 venting wells) Average venting time, hours 0.0833 2.99 2.6 0.5425 0.5 Average gas flow rate, Mscfd 250 727 875 Average applied 1500 1,276.8 Total emissions, scf gas/yr 101,698,021 124,984 906,144 529,679 66,812 Emissions per well, scfy gas/well 2,118,709 31,246 14,158 18,265 3,712Summary and Analysis of API and ANGA Survey Responses 45

52. TABLE C-4. LIQUIDS UNLOADING FOR UNCONVENTIONAL GAS WELLS WITH PLUNGER LIFTS, CONTINUED NEMS Region Rocky Mountain # venting gas wells 247 23 296 19 793 # gas well vents 1,476 51.43 2,080 21,888 9,516 Average tubing diameter, inches 1.997 1.92 2.375 2.375 2.375 Average well depth, feet 11,149 11,164 11,056 10,844 7,400 Average surface pressure, psig 250 290 250 198 150 (for venting wells) Average venting time, hours 0.407 1.12 2.1 0.455 0.67 Average gas flow rate, Mscfd 127 454 433 83 46 Total emissions, scf gas/yr 6,070,440 238,833 12,027,460 98,082,094 22,045,130 Emissions per well, scfy gas/well 24,577 10,384 40,633 5,162,215 27,800Summary and Analysis of API and ANGA Survey Responses 46

53. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASEDThe calculated emissions shown in Tables C-1 through C-4 are based on applying Equation W-8from 40 CFR 98 Subpart W to gas well liquid unloading without plunger lifts and Equation W-9to gas well liquid unloading with plunger lifts. The equations and the terms are provided below. 98.233(f)(2) Calculation Methodology 2. Calculate the total emissions for well venting for liquids unloading using Equation W–8 of this section. Where: E s,n = Annual natural gas emissions at standard conditions, in cubic feet/year. W= Total number of wells with well venting for liquids unloading for each sub-basin. −3 0.37×10 = {3.14 (pi)/4}/{14.7*144} (psia converted to pounds per square feet). CD p = Casing internal diameter for each well, p, in inches. WD p = Well depth from either the top of the well or the lowest packer to the bottom of the well, for each well, p, in feet. SP p = Shut-in pressure or surface pressure for wells with tubing production and no packers or casing pressure for each well, p, in pounds per square inch absolute (psia) or casing-to-tubing pressure of one well from the same sub-basin multiplied by the tubing pressure of each well, p, in the sub-basin, in pounds per square inch absolute (psia). Vp= Number of vents per year per well, p. SFR p = Average flow-line rate of gas for well, p, at standard conditions in cubic feet per hour. Use Equation W–33 to calculate the average flow-line rate at standard conditions. HR p,q = Hours that each well, p, was left open to the atmosphere during unloading, q. 1.0 = Hours for average well to blowdown casing volume at shut-in pressure. Z p,q = If HR p,q is less than 1.0 then Z p,q is equal to 0. If HR p,q is greater than or equal to 1.0 then Z p,q is equal to 1. 98.233(f)(3) Calculation Methodology 3. Calculate emissions from each well venting to the atmosphere for liquids unloading with plunger lift assist using Equation W–9 of this section. Where: E s,n = Annual natural gas emissions at standard conditions, in cubic feet/year. W= Total number of wells with well venting for liquids unloading for each sub-basin. −3 0.37×10 = {3.14 (pi)/4}/{14.7*144} (psia converted to pounds per square feet). TD p = Tubing internal diameter for each well, p, in inches. WD p = Tubing depth to plunger bumper for each well, p, in feet. SP p = Flow-line pressure for each well, p, in pounds per square inch absolute (psia), using engineering estimate based on best available data. Vp= Number of vents per year for each well, p. SFR p = Average flow-line rate of gas for well, p, at standard conditions in cubic feet per hour. Use Equation W–33 to calculate the average flow-line rate at standard conditions. HR p,q = Hours that each well, p, was left open to the atmosphere during each unloading, q. 0.5 = Hours for average well to blowdown tubing volume at flow-line pressure.Summary and Analysis of API and ANGA Survey Responses 47

54. DO NOT DISTRIBUTE UNTIL FORMALLY RELEASED Z p,q = If HR p,q is less than 0.5 then Z p,q is equal to 0. If HR p,q is greater than or equal to 0.5 then Z p,q is equal to 1.Summary and Analysis of API and ANGA Survey Responses 48

SlideShare for iOS

Views

Actions

Embeds 80

Report content

Transcript