LETTER • OPEN ACCESS

Emissions in the stream: estimating thegreenhouse gas impacts of an oil and gas boomTo cite this article: Andrew R Waxman et al 2020 Environ. Res. Lett. 15 014004

 

View the article online for updates and enhancements.

You may also likeMathematical Model of the Water Qualityin Kalibaru WatershedB Hariono, R Wijaya, M F Kurnianto et al.

-

Assessments of Heavy Metal Zn andColiform in Midstream of Blanakan River,Subang, West JavaT Damayanti and N D Takarina

-

Flood control and loss estimation forpaddy field at midstream of Chao PhrayaRiver Basin, ThailandT C Cham and Y Mitani

-

Recent citationsThe hidden costs of energy and mobility: Aglobal meta-analysis and researchsynthesis of electricity and transportexternalitiesBenjamin K. Sovacool et al

-

Electrofuels from excess renewableelectricity at high variable renewableshares: cost, greenhouse gas abatement,carbon use and competitionMarkus Millinger et al

-

Interactions of CO2 with HydrocarbonLiquid Observed from Adsorption of CO2in Organic-Rich ShaleJinsheng Wang et al

-

This content was downloaded from IP address 65.21.228.167 on 20/10/2021 at 18:30

Environ. Res. Lett. 15 (2020) 014004 https://doi.org/10.1088/1748-9326/ab5e6f

LETTER

Emissions in the stream: estimating the greenhouse gas impacts ofan oil and gas boom

AndrewRWaxman1 , AchmadKhomaini2, BenjaminDLeibowicz3 and SheilaMOlmstead1,4

1 LBJ School of Public Affairs, University of Texas at Austin, Austin, TX,United States of America2 Jackson School of Geosciences, University of Texas at Austin, Austin, TX,United States of America3 Operations Research and Industrial Engineering, University of Texas at Austin, Austin, TX,United States of America4 Resources for the Future,Washington, DC,United States of America

E-mail: awaxman@utexas.edu

Keywords: oil and gas, shale, GHGemissions, industrial emissions, gulf region

Supplementarymaterial for this article is available online

AbstractThe Shale Revolution has stimulated a large and rapid buildout of oil and gas infrastructure in theGulfand Southwest regions of theUnited States (US), expected to unfold over decades. Therefore, it iscritical to develop a clearer understanding of the scale and composition of the likely greenhouse gas(GHG) emissions associatedwith this activity.We compile a detailed inventory of projected upstreamoil and gas production expansions aswell as recently and soon-to-be builtmidstream and downstreamfacilities within the region.Using data from emissions permits, emissions factors, and facilitycapacities, we estimate expectedGHGemissions at the facility level for facilities that have recentlybeen constructed or are soon to be constructed. Our central estimate suggests that the total annualemissions impact of the regional oil and gas infrastructure buildoutmay reach 541million tons of CO2

equivalent (CO2e) by 2030, which ismore than 8%of total USGHGemissions in 2017 and roughlyequivalent to the emissions of 131 coal-fired power plants. A substantial fraction of the projectedemissions come frompetrochemical facilities (38%) and liquefied natural gas terminals (19%).Researchers have largely focused on upstream emissions such as fugitivemethane (CH4) associatedwith newUS production; ourfindings reveal the potentially greater prominence ofmidstream anddownstream sources in the studied region.

1. Introduction

The Shale Revolution has transformed North Amer-ican and global energy markets. It has reduced energyprices (Hausman and Kellogg 2015, Kilian 2016);shifted electricity production from coal toward naturalgas, reducing local and regional air pollution (Johnsenet al 2019); provided a source of employment andincome during the Great Recession (Feyrer et al 2017);and allowed the US to approach the point at which itexports more energy than it imports (EIA 2018)5.

There have also been harmful impacts (Olmstead et al2013, Graham et al 2015, Hill and Ma 2017, KomarekandCseh 2017).

This study quantifies the long-run greenhouse gas(GHG) impacts of the ongoing buildout of oil and gasinfrastructure in the US Gulf region including on- andoffshore facilities in Texas (TX) and Louisiana (LA). Werefer to the geographic scope of the study as the Gulfeven though it includes plays in TX that extend into theSouthwest6. Our scope comprises portions of the oil andgas value chain from production (upstream), to trans-mission and distribution (midstream), to industrial end

OPEN ACCESS

RECEIVED

22August 2019

REVISED

2December 2019

ACCEPTED FOR PUBLICATION

3December 2019

PUBLISHED

14 January 2020

Original content from thisworkmay be used underthe terms of the CreativeCommonsAttribution 3.0licence.

Any further distribution ofthis workmustmaintainattribution to theauthor(s) and the title ofthework, journal citationandDOI.

5A controversy exists about the extent towhichCH4 emissions from

natural gas extraction could mean that GHG emissions increaserather than decrease from switching from coal to natural gas as anenergy source (Howarth et al 2011, Wigley 2011). However, someresearch suggests this effect may be overstated (Cathles et al 2012,Jenner and Lamadrid 2013).

6The scope includes upstream activities in TX and LA (including

the Permian, Barnett, and Eagle Ford areas) and offshore in thebroader Gulf of Mexico, and midstream and downstream facilitiesin TX and LA.

© 2020TheAuthor(s). Published by IOPPublishing Ltd

uses (downstream)7. Because of the path dependency ofindustrial development andhigh adjustment costs of fuelswitching, further transition toward oil and gas maymean larger future use of fossil fuels relative to renew-ables (Unruh 2000). Therefore, it is critical to develop aclearer understanding of the scale and composition oftheGHG impact of these newoil and gas resources.

Facility-level research on industrial GHG emis-sions is limited (Hamit-Haggar 2012, Ryan 2012). Incontrast, methane (CH4) emissions from upstreamoil and gas production have received significantattention (Alvarez et al 2012, Howarth 2014, Elvidgeet al 2018). This study accounts for emissions pro-duced by the industrial development that an oil andgas boom stimulates in downstream sectors such aspetrochemicals and refining, as well as those from oiland gas production. Rather than retrospectivelyexamine a past buildout of oil and gas infrastructureor use a model to simulate future development, weinventory recently or soon-to-be built facilities fromgovernment, industry, and media sources. This pro-vides a detailed picture of expected emissions fromnew facilities and other infrastructure, based on datafrom the regulators who permit these facilities andthe industries that invest in them.

Results suggest that the total annual emissionsimpact of the oil and gas infrastructure buildout in theGulf may reach 541 million tons of carbon dioxideequivalent (CO2e) by 2030, more than 8% of total USGHG emissions in 2017 (EPA 2019a). While impactson upstream emissions such as fugitive CH4 are dis-cussed in the literature, our findings reveal the poten-tially greater prominence of mid- and downstreamsources in some areas.

1.1. Trends in oil and gas development in the gulfregionThe earliest TX oil discoveries date to the 1860s.Discoveries offshore in Galveston Bay followed severalyears later, and the region has continued to developsince. Figure S1 in the supplemental information (SI)shows the evolution of oil and gas production in theregion and for the US as a whole since 1996. Trendssince the mid-2000s reflect increases in productionfrom the Permian Basin, Eagle Ford Play, Barnett Play,and TX-LA-Mississippi Gulf Basin (also called the SaltBasin) as well as those elsewhere in the US such as theBakken andMarcellus.

A major result of the recent boom is the transitionaway from coal and towards natural gas in electricitygeneration, which has lowered the carbon intensity ofelectricity in many parts of the US. However, thebroader impacts on GHG emissions of new gas

production are uncertain, particularly if natural gasprices remain low (Chen et al 2019) and/or fugitiveCH4 is high (OGJ 2019). The boom in oil productionin this region and elsewhere is likely to increase GHGemissions both indirectly, if it puts downward pres-sure on global oil prices, and directly due to emissionsfrom the upstream, midstream, and downstreamactivities we describe in this paper.

1.2. Relation to the existing literatureThe ongoing expansion of production in the Gulf andSouthwest regions, particularly in the Permian Basin,has been described in the literature (Wallace 2019)8.Newell and Raimi (2014) use the EPA Inventory ofGHG Emissions and Sinks and EIA National EnergyModeling System (NEMS) data to predict a smallincrease in GHGs by 2040 from new production,depending on oil prices, extraction technology, andthe global warming potential (GWP) of CH4. Haus-man and Kellogg (2015) predict damages fromincreased emissions of $2.5–15 billion per year. Bothstudies predated awareness of the full extent of oiland gas resources in the Permian (Gaswirth et al2018). Newell and Raimi (2014) project emissionsbased on oil and gas production, using emissionsfactors that reflect average mid- and downstreamemissions. They do not project emissions at thefacility level. In addition, the focus of that study isnational, including impacts through fuel substitutionand trade, as well as those from the use of oil and gasfor electricity and transportation, not consid-ered here.

A large engineering literature models climateimpacts of the fossil fuel sector (DeLuchi 1993, Daviset al 2010, Alvarez et al 2012, Burnham et al 2012,Allen et al 2013, Subramanian et al 2015, Zimmerleet al 2015, Alvarez et al 2018, Tong et al 2019). In con-trast to the current approach, these studies applymodeling assumptions about future oil and gas infra-structure and then extrapolate expected emissionsfrom a simulated buildout. This literature has beenextremely helpful for characterizing the largest sour-ces of future GHG emissions.

The approach taken here addresses a gap in the lit-erature, examining individual facilities that have actu-ally been built or will soon be built, and estimatingtheir associated emissions. We focus on the most den-sely developed region of oil, gas, and petrochemicalinfrastructure in the US. However, given that thesecommodities are transported long distances and tra-ded in internationalmarkets, where end uses and asso-ciated carbon intensity may depend critically on

7The upstream scope includes production expansions from new

and existing wells. Industrial downstream uses considered hereinclude a variety of petrochemical activities described in thesupplemental information (SI) table S1 available online at stacks.iop.org/ERL/15/014004/mmedia.

8This includes the Wolfcamp Shale and Bone Spring Formation,

estimated in November 2018 to be collectively the largest contin-uous oil and gas resource ever identified (Gaswirth et al 2018). Theexpansion of production continues, though rig counts have at leasttemporarily fallen (Blum2019).

2

Environ. Res. Lett. 15 (2020) 014004

market conditions, we restrict the focus of our analysisto a subset of activities described in section 2.

Alvarez et al (2018) corroborate bottom-up esti-mates of existing facility-level emissions with top-down observations and find US CH4 emissionsbetween 8 400 and 15 000 Gg per year. Our estimatesof CH4 emissions are consistent with these in thatthey do not exceed their total emissions numbers, butwe only account for a small subset of oil and gasproduction considered in that study, so the scope isdifferent. Tanaka et al (2019) perform LCA for thecoal-to-gas shift to determine the climate impacts.Estimates of CO2 and CH4 emissions from our studyare not larger than theirs, but again, their analysis isnational, so the comparison cannot tell us whetherthe regional estimates in this paper are consistentwith their results.

Other studies project the future inventory ofindustrial facilities using current sector-level emis-sions estimates along with assumptions about futurechanges in the age distribution of facilities (Davis et al2010, Tong et al 2019). While these assumptionsenable global forecasts, they necessarily extrapolatefrom one sector or dataset9. Our inventory approachuses estimates of actual facility-level production andemissions, compiled from regulatory agencies, indus-try reports, and other primary sources, rather thansimulating emissions using models that aggregatefacilities sectorally, nationally, or globally. The resultsfromour granular approach can be used to validate theemissions projections from more aggregated models.In addition, identifying the sources of significantpotential emissions growth is critical for formulatingeffectivemitigation strategies.

The approach taken here is generally one describedin the engineering literature as ‘bottom-up’ in that weapply emissions factors facility-by-facility and aggre-gate. Other approaches use aircraft or satellitemeasure-ments of regions to produce ‘top-down’ estimates.Comparisons of the two approaches suggest that the‘bottom-up’ approach understates total emissions inmany cases (Allen 2016).

The expansion of the oil and gas sector in the Gulfregion has reduced the cost of natural gas and petro-leum and, in turn, lowered the cost of producing pet-rochemical products at scale (EIP 2018). Cumulativepetrochemical investments related to unconven-tional oil and gas are estimated at $204 billion fromDecember 2010 to May 2019 (ACC 2019b). Whatcould account for this massive buildout?A lower

natural gas price means greater availability of petro-chemical feedstocks (e.g. ethane from natural gasliquids). It also means that the cost of energy used tofuel conversion of feedstocks into refined products(e.g. ethane crackers that produce ethylene) hasfallen. This is borne out in market conditions forethylene and other petrochemical derivatives, whereprices are falling and production is growing(Hay 2019). All of this increases the supplies of theseproducts and their corresponding GHG emissions(SI figure S2).

2.Data andmethodology

We build a detailed inventory of facilities using datafrom a variety of sources. Our analysis of themidstreamand downstream segments of the value chain operatesat the facility level, where the inventory tracks construc-tion of new facilities and expansion of existing facilities.The timing of planned or under-construction facilitiesis difficult to ascertain and subject to change. Further-more, facilities not yet announced (and thus excludedfrom our inventory) may become operational prior tothe completion of some facilities we do observe. There-fore, the timeframe for considering newmidstream anddownstream facilities naturally corresponds to theannouncement/planning, approval, and constructiontimelines of the facilities for which information isavailable. Initial data collected provide the location,production type, and in most cases the expectedproduction capacity of each facility.

Our classification of facilities as down- versus mid-stream is basedonwhere they fall in the regional produc-tion process, which may differ from studies with anational or global focus. Refining of oil and gas productsinto fuels or chemical products (e.g. ethane cracking,which is classified as ‘petrochemical’ in our analysis) isclassified as downstreamproduction. Natural gas and oilstorage and export, however, are classified asmidstream,with the understanding that these products will even-tually beused in somedownstreamprocess elsewhere.

For facilities where emissions permits are alreadyapproved, we use permit data to quantify expectedemissions. For those without permits, we use volu-metric emissions factors to predict what emissionswill likely be given the production activity and repor-ted facility capacity. Permitted emissions may over-(or under-) state actual emissions to the extent thatindividual facilities fall below or exceed permittedlevels. Emission factors may also deviate from actualemissions if individual facility production processesemit more or less than the average facility in our data.For upstream oil and gas production, we track pro-jections of future development to 2030 and assumethat the mid- and downstream projects in our inven-tory are fully operational by that date. Accidentsrelease large amounts of GHGs and may not be cap-tured in either approach, which would mean we are

9In Tong et al (2019), the authors construct emission estimates

‘assuming that the age distribution and survival curves of eachregion’s industry infrastructure are consistent with its electricityinfrastructure’ (p 378). In Davis et al (2010), the authors acknowl-edge that they ‘make the arbitrary assumption that CARMA’semissions and energy data for 2009 (or, occasionally, 2004) are anaccurate estimate throughout a plant’s lifetime (despite evidence tothe contrary)’ (p 2), referring to their use of the Carbon Monitoringfor Action (CARMA) database.

3

Environ. Res. Lett. 15 (2020) 014004

understating emissions10. Section 3.4 discusses sensi-tivity analysis related to some of these sources ofuncertainty.

2.1.Data sources2.1.1. Inventory of new facilitiesTable 1 groups the components of the present analysisinto segments of the regional oil and gas stream. Weseparate compression, boosting, fractionization, andother activities from upstream oil and gas production,because we observe new facilities and infrastructurerelated to these activities, whereas we rely on produc-tion projections to predict upstream (well-level) emis-sions. We also include fuel terminals, refining, andpetrochemicals in the downstream segment based onthe set of activities covered in our analysis, acknowl-edging that in other frameworks these may be mid-stream activities11.

Table 2 presents a summary of the 327 mid- anddownstream facilities in our data. New facilityannouncements and their associated production capa-cities come from company websites, industry andmedia sources, the American Chemistry Council, andenvironmental advocacy websites (BMI/Fitch 2018,Fitch 2018, ACC 2019a, EIP 2019)12. Where permitsexist, permitted emissions levels come from an onlinedatabase of Federal Energy Regulatory Commission,Environmental Protection Agency (EPA), TX Com-mission on Environmental Quality (TCEQ), and LADepartment of Environmental Quality regulatorydocuments13. The classification of facilities into seg-ments of the value chain is not always straightforward.One comparator is the OPGEE model (El-Houjeiriet al 2018), which covers activities in the stream up tobut not including refining, and is distinct from ourown framing of the oil and gas value chain. Figure 1illustrates the share of each facility type in the inven-tory at each development stage (planned, under con-struction, completed, and unknown).

2.1.2. Upstream productionProjections of upstream oil and gas production in theGulf region are obtained from the EIA Annual EnergyOutlook (EIA 2019). These projections are based onNEMS scenario results, combining data on domesticand international supply and demand factors andeconomic forecasts to predict the evolution of oil and

gas prices and quantities. We consider projectionsfrom several differentNEMS scenarios.

2.2. Approach for estimating emissionsTable 3 reports the emissions factors applied in ouranalysis. SI figure S3 provides a visual summary of theemissions factor estimation process, and how it differsfor upstream versus downstream and permitted versusunpermitted facilities. Where facility permits includeemissions, we use these data to construct our esti-mates. Where facility-level permitted emissions areunavailable, we employ a variety of approaches toobtain emissions factors that represent GHG emis-sions per unit of output. An alternative approachwould be to use energy input emissions factors andmultiply these by facility-level fuel use. Unfortunately,we cannot reliably obtain fuel specifications andprojected fuel use for the facilities in our data14. Thisalsomeans thatwe donot capture emissions associatedwith off-site electricity use or diesel from railcars orother oil and gas transportation. CH4 emissions areconverted to CO2e based on 100 year GWP from theIntergovernmental Panel on Climate Change (IPCC),with a value of 32 for CH4; alternative GWPs areconsidered in section 3.1. We ignore emissions offluorinated gases and N2O. These high-GWP gases areharder to forecast by facility and constitute a smallshare ofGWP from the oil and gas sector.

2.2.1. Upstream emissions factorsFor upstream production, we cannot disentangleproduction forecasts by drilling method, so we ignorenon-CH4 emissions and focus exclusively on vented,flared, or fugitive CH4 released during production,using estimates fromBrandt et al (2014) for natural gasand Cai et al (2014) for oil15. Again, note that we donot include the combustion of fuels for energy at wellsites or farther downstream (say, for power genera-tion), acknowledging that they contribute the bulk ofeventual downstream emissions from these fuelsources16.

10Brandt et al (2014) show that differences between bottom-up and

top-down emissions estimates may be explained to a large extent bythese events, which are not incorporated in the latter (whichincludes EPA’s GHGReporting Program).11

In table S1, we itemize the set of facility types classified asPetrochemicals in our data. Further information about dataconstruction can be found in the SI, section B.12

All sources used to construct these data are provided in SI tablesS2 and S3.13

These documents are publicly available from the EnvironmentalIntegrity Project in an online database (EIP 2019).

14A common approach in the energy engineering literature applies

emissions factors from lifecycle analysis (LCA), either based onvolumes of fossil fuel extraction or end use. This study does not useLCA-based emission factors as they would likely lead to doublecounting of emissions atmultiple points of the value chain.15

These factors are based on volumetric leak, vented, and flaredemissions from CO2 and CH4 reported in the supplemental data ofBrandt et al (2014). We test the robustness of emissions estimates tovariation in these upstream emission factors by considering analternative set of factors based onAlvarez et al (2018). This sensitivityexercise is discussed in more detail in section 3.4 and in thesupplemental information.16

A referee pointed out the importance of CO2 emissions for oilwells with enhanced oil recovery (Brandt et al 2018, Masnadi et al2018). While these emissions are critical in evaluating crudeproduction nationally or globally, the basins in this study have notyet applied these techniques on a meaningful scale (Du andNojabaei 2019).

4

Environ. Res. Lett. 15 (2020) 014004

2.2.2.Mid- and downstream emissions factorsWhere facilities in our inventory have already beenpermitted by the EPA,we obtain facility-level emissionsestimates (EPA 2019b). When this is not the case, weadopt two alternative approaches. Where we havepermit data for some facilities of a given type, wecalculate the average CO2e emissions per unit of outputacross permitted facilities of that type, and use thataverage factor for any facilities without permit data17.For facility types lacking permitted emissions data foreven a single facility, we use emissions factors from theliterature. Factors for petrochemical facilities are morechallenging to construct because they vary dependingupon the particular processes employed and chemicalsproduced.We construct amore refined set of factors forthese facilities by facility process type in SI table S118.

Given the literature’s strong focus on upstreamoil and gas emissions, why should petrochemicalsaccount for such a substantial share of total emis-sions in our analysis?Among the highest emittingfacilities in our data, the largest proportion are LNGterminals (19%), ethane crackers (13%), and thoseproducing polyethylene and ethylene derivatives(9%), methanol (6%), and fertilizers (5%). This isconsistent with lower costs of natural gas extractiondriving growth in domestic and export-oriented pro-duction of precursors to plastic production, metha-nol for fuel and plastics, and fertilizers foragriculture. Moreover, the large share of LNG term-inal emissions in our database reflects the extent towhich the saturation of low-cost natural gas in the US

market is increasing exports. Seven of the eight lar-gest US refineries (EIA 2019) and the vast majority ofcurrent and expected future US LNG export capacity(EIA 2018) are on the TX and LA Gulf coasts. Theregion contains the ‘highest collective concentrationof petroleum refining and petrochemical productioncapacity of just about anywhere in the world’(Dismukes et al 2019). In this setting, downstreamemissions likely represent a substantial share of thetotal.

We do not estimate emissions for 17 inventoriedfacilities. Six of these projects are on hold, so emissionsmay not materialize. Two are transboundary pipelineswith themajority inMexico, outside of our scope. Theremaining nine omitted facilities are eight petrochem-ical facilities and one natural gas liquids terminal forwhich we are unable to obtain relevant emission fac-tors. A complete description of data sources is pro-vided in the SI (tables S2 and S3).

3. Results

3.1. Upstream emissionsFigure 2 presents projections of oil and gas productionbased on EIA NEMS data for 2017–203019. Panel Ashows oil and gas production by region for thereference case scenario, a business-as-usual (BAU)forecast of economic and market conditions20. Thefigure reflects the growing importance of Permian oiland gas as well as onshore Gulf Coast (Salt Basin) andEagle Ford resources. The large difference between gasand oil emissions reflects the fact that a larger share of

Table 1.Oil, gas, and petrochemical activities included and not included in this study.

Upstream Midstream Downstream

Included Petroleum Exploration abd development*,

Production inside the Southwest

andGulf Basins*

Transport Export terminals*, Refining, Petrochemical

production

Natural

Gas

Processing*,

Compression*,

Storage*,

Transport*

Natural

Gas

Liquids

Fractionation,

Transport

Not included Production outside of Southwest

andGulf Basins, emissions from

electricity use/combustion at

well level

Electricity gen-

eration/com-

bustion done

off-site

Electricity generation/combustion done off-

site, combustion or fugitive emissions from

end-uses* (e.g. from gasoline for

transportation)

Note.Processes with an asterisk (*) includeCH4 emissions fromventing, flaring (alongwithCO2 emissions), or fugitive sources.

17Before calculating the emission factor based on the average for

each facility type, we remove outliers in the relationship betweenproduction capacity and emissions.18

For three petrochemical facility types lacking permit data, we useLCA-based emission factors from ecoinvent (Frischknecht et al2005). These facilities account for a very small amount of totalemissions in our analysis (157 thousand tons of CO2e per year, or0.03%of total emissions).

19Production for 2017–2018 is actual rather than projected, but is

included for illustrative purposes.20

Onshore Gulf production includes coastal TX, LA, Mississippi,Alabama, and Florida (dominated by production in the first twostates). Southwest production includes western TX and eastern NewMexico.

5

Environ. Res. Lett. 15 (2020) 014004

gas emissions occurs further upstream in the valuechain than for oil21.

Figure 2, panel (B) demonstrates the variation inprojected emissions for four alternative NEMS scenar-ios: high and low expected oil prices, and high and low

resource extraction and technology costs. These casesproduce a range of 140–200 million tons of expectedCO2e emissions annually by 2030 from upstream pro-duction alone. We can also bound these estimates byadjusting the assumedGWP to the lower value used bythe US EPA or the higher 20 year IPCC value(IPCC 2014). Assuming aGWP for CH4 of 86, the highemissions case rises to 531 million tons of CO2e. With

Table 2.Number of facilities by Facility Type AndProject Status.

Planned Construction Completed Unknown

Total

in TX

Total TX

and LA

%ofGrand

Total

Midstream 117 37 4 1 118 159 48.62%

Compressor station 12 0 0 1 3 13 3.98%

Gas Processing 18 2 0 0 17 20 6.12%

Pipeline, Gas 22 18 1 0 28 41 12.54%

Pipeline,NGL 11 4 2 0 17 17 5.20%

Pipeline,Oil 24 10 0 0 31 34 10.40%

Rail Terminal 1 1 0 0 2 2 0.61%

Storage, Gas 1 1 0 0 2 2 0.61%

Terminal, Oil 12 0 0 0 11 12 3.67%

Terminal, LNG 16 1 1 0 7 18 5.50%

Downstream 84 43 34 7 112 168 51.38%

Fractionator 9 3 0 0 11 12 3.67%

Petrochemical 65 23 34 7 81 129 39.45%

Refinery 10 17 0 0 20 27 8.26%

Total in TX (Mid- and

Downstream)140 61 23 6 230

GrandTotal 201 80 38 8 230 327

Figure 1.Estimated annual emissions by facility type and project completion status.Notes: results in the figure do not includeemissions for 17 facilities. For six of them the project is on hold, two are transboundary pipelines, two are pipelines with insufficientdata to approximate emissions, and the remaining nine are petrochemical facilities and anNGL storage facility for whichwe areunable to estimate emissions.

21We report projected oil and gas production for different parts of

theGulf and Southwest regions in SIfigure S4.

6

Environ. Res. Lett. 15 (2020) 014004

Table 3. Facility-level emissions factors by facility type.

Category Facility type Value Unit Source

Upstream Gas (CH4) 6.63×106 Mil Ton of CO2e/Mil cubic feet of gas Brandt et al (2014)Upstream Oil (CH4) 3.58×10−9 Mil Ton of CO2e/MMBtu Cai et al (2014)Midstream Compressor station (CO2) 0.0960 Mil TonPer Year of CO2e/bcfd Averaged fromPermitData

Midstream Gas Processing 0.898 Mil TonPer Year of CO2e/bcfd Averaged fromPermitData

Midstream Pipeline, NGL (CH4) 3.59×10−5 Mil Ton of CO2e/Mi EPA (2017b)Midstream Pipeline, Oil (CH4) 1.73×10−7 Mil Ton of CO2e/Thousand cubicmeters of oil Picard (1999)Midstream Rail Terminal (CH4) 8.00×10−7 Mil Ton of CO2e/Thousand cubicmeters of oil Picard (1999)Midstream Rail Terminal (CO2) 2.30×10−9 Mil Ton of CO2e/Thousand cubicmeters of oil Picard (1999)Midstream Storage, gas (CH4) 2.42×10−5 Mil Ton of CO2e/Mil cubicmeters of gas Picard (1999)Midstream Terminal, LNG 2.25 Mil TonPer Year of CO2e/bcfd Averaged fromPermitData

Midstream Terminal, Oil 3.60×10−8 Mil TonPer Year of CO2e/bpd Averaged fromPermitData

Downstream Fractionator 2.42×10−6 Mil TonPer Year of CO2e/bpd Averaged fromPermitData

Downstream Refinery 1.65×10−5 Mil TonPer Year of CO2e/bpd Averaged fromPermitData

Downstream Petrochemical Depends on Process, see SI table A1.

Note. Emissions factors are based on observed unit CO2e emissions per unit of output where permit data is available as denoted by ‘Averaged fromPermit Data.’Where not available, factors have been obtained from technical documents or

research papers cited in the rightmost column. 100 year global warming potential of CH4 is assumed to be 32 following EPA (2017a).

7

Environ.R

es.Lett.15(2020)014004

the low-case GWP of 25, the low emissions case falls to108 million tons of CO2e. Assuming our baselineGWP for CH4 of 32, our results then predict anincrease in annual emissions from 2017 to 2030 ran-ging between 34–99 million tons of CO2e. The refer-ence case, used below for aggregate emissions, is anincrease of 67.5million tons of CO2e

22.

3.2.Midstream anddownstream emissionsTable 4 reports estimates of total emissions by facilitytype and project completion status. By a simple countfrom table 2, petrochemical plants and pipelinescomprise the largest numbers of new facilities, reflect-ing, in part, variation in the scale of activity per facilityby type (e.g. one new refinery constitutes a largerelative increase in total refining capacity in theregion). Mid- and downstream emissions are expectedto be 371.7million tons of CO2e per year by 2030.

Table 4 shows the large impact of petrochemicalemissions, totaling 206.3 million tons of CO2eannually, as well as that of downstream emissions as awhole (250.3 million tons). While we cannot pinpointthe exact timing of the construction of these facilities,22% of expected downstream emissions come fromfacilities already completed or under construction,74% from those planned, and only 4% from facilitieswherewe are unable to ascertain project status.

3.3. Aggregate emissionsIf all of the projects in table 2 come to fruition by 2030,total emissions (including upstream) would increase byabout 541.1 million tons of CO2e annually by that yearbased on the reference case scenario. Assuming a GWPforCH4of 32, upstreamemissions increases could vary by65 million tons depending upon the relevant scenario.Assuming that projects currently under constructioncome online in 2021, and that planned projects and thosewith unknown status come online in 2024, total cumula-tive emissions from2017 to2030wouldbe2.6billion tonsCO2e

22. With a discount rate of 3% and a global socialcost of carbon ranging between $39 and $50/ton (Inter-agency Working Group 2016), this would imply addi-tional external damages just fromGHGemissions of $112billion, relative to2017 emissions.

3.4. Uncertainty analysisIn SI section A, a sensitivity analysis explores how ourestimates varywith key input assumptions.Our approachimplies potential ranges of upstream emissions between138and200million tonsofCO2eper year, andapotentialreduction of our central estimate for mid- and

downstream facilities from 371.7 million tons of CO2eper year down to 142 million if planned facilities do notmaterialize23. Additionally, because actual emissionsmayvary from permitted emissions, we use reported emis-sions from EPA FLIGHT to characterize expected varia-tion in realized emission rates. We find that whilepermitted and reported emissions per facility are similaron average, accounting for this variation could reduceourestimate of total emissions to 374 million tons of CO2eper year. Thus, even the most conservative assumptionsaboutupstreamoil andgasproduction, the completionofplanned mid- and downstream facilities, and permittedversus actual emissions imply a large GHG emissionscontributionof the regional buildout.

4.Discussion and conclusions

Weconstruct an inventoryof recentlyor soon-to-bebuiltfacilities that are part of the rapid buildout of oil and gasinfrastructure in the US Gulf region. Expected GHGemissions from each facility are obtained directly fromemissions permits or indirectly by deriving emissionsfactors and multiplying them by anticipated facilitycapacity. This approach enables us to assess the scale ofthe aggregate GHG emissions impact of the oil and gasboom in the region and decompose it by value chainsegment, facility type, andproject completion status.

Our main estimate suggests that the buildout willincrease annual GHG emissions by about 541 milliontons of CO2e by 2030, amounting to more than 8% oftotal US GHG emissions in 2017 (EPA 2019a). Forcontext, this is roughly equivalent to the CO2e emis-sions of 131 coal-fired power plants, or 82% of Texas’total 2016 GHG emissions. The downstream segmentof the buildout—especially petrochemical facilities,which account for more than one-third of estimatedaggregate emissions—is a major contributor. LNGterminals, in the midstream segment, comprise morethan one-fifth of the total. While prior work highlightsupstream emissions such as fugitive CH4, our findingssuggest that the mid- and downstream infrastructurebuildouts stimulated by the oil and gas productionexpansion may have comparable or greater GHGimpacts in this region. To put this into perspective, thebuildout documented here suggests potential mid-and downstream emissions will be 384.4 million tonsof CO2e, which is 7% larger than total US industrialemissions in 2017 (EPA2019a).

22How to apply GWPs to emission estimates is an open question,

with alternative approaches discussed by Allen et al (2016) andTanaka et al (2019).22

Because we are unable to explicitly link the different productionexpansion scenarios to resulting changes in the levels of mid- anddownstream activities, we cannot consider the extent to whichalternative upstream scenarios might alter mid- and downstreamemissions.

23Upstream emissions sensitivity is based on EIA production

scenarios based on the high and low cases in the trajectory of oilprices and oil and gas technology’s effect on production. We alsoconsider an alternative set of emission factors based upon Alvarezet al (2018), and find that it increases natural gas emissions byroughly an order of magnitude (325 million tons CO2e per year by2030) and decreases oil emissions by a smaller proportion (16million tons CO2e per year by 2030). These factors may not be asappropriate as those used in our baseline estimates because theproduction sites only include one of the plays (Barnett Shale) in ouranalysis.

8

Environ. Res. Lett. 15 (2020) 014004

More and better data (e.g. emissions factors) are nee-ded to enable future research projectingmid- and down-stream emissions and developing mitigation strategies.Only about 30% of the facilities in our database arealready operating or under construction, thus our resultscan inform decision-making on policy and technologystrategies for planned facilities. Our estimates assume aBAU scenario for the emissions intensities of industrialprocesses. Through a variety of voluntary or regulatoryapproaches, emissions could be lower. While we focuson GHG emissions, the regional human health implica-tions of associated emissions of local pollutants, such asparticulate matter and ozone precursors, could havehigher economicdamages.

Should steps be taken to reduce the buildout wedescribe in the Gulf region, because all of the relevantmarkets cross regional and national boundaries, muchof that buildoutmight occur elsewhere. Thus, the coun-terfactual to the buildout investigated here could involveeither higher or lower emissions, depending on howmuch of the buildout would be displaced, and where itwould locate.

Our approach has several limitations. Mostimportantly, we limit our scope to the oil and gas valuechain within theGulf region in theUS.We do not con-sider the GHG impacts of fuel use and substitution inother end uses (e.g. residential, commercial, elec-tricity, transportation) or changes in oil and gas trade

Figure 2.Emissions Estimates by Region and EIA Scenario.Notes: panel (A) graphs annual production estimates based onEIAreference scenario. Panel (B) includes production for EIA Southwest, GulfOnshore, andGulf Offshore regions. Emissions estimatesare based on emissions factors fromAlvarez et al (2018). Panel (B) emissions estimates based on five separate production scenarios forSouthwest, Gulf Onshore, andGulfOffshore regions. OnshoreGulf production corresponds to coastal TX, LA,Mississippi, Alabama,and Florida. Southwest production includes westernTX and easternNewMexico.

9

Environ. Res. Lett. 15 (2020) 014004

resulting from expanded regional production. Wefocus on permitted emissions or those based on aver-age emissions factors, which may over- or understateemissions. Facilities’ realized emissions may differfrom permitted levels. Rough sensitivity analysisaround major sources of uncertainty suggests thateven lower-bound estimates of the incremental emis-sions from the regional buildout are large.

Acknowledgments

We acknowledge financial support from the Cynthiaand GeorgeMitchell Foundation. Findings, methods,and results are completely independent from theFoundation. We thank Matias Navarro and SeanCorcoran for research assistance and Sarah Brennanfor an initial inventory of facilities and sources.Daniel Raimi provided helpful comments on a draftof this paper.

Data availability statement

The data that support the findings of this study areavailable from the corresponding author upon reason-able request.

ORCID iDs

AndrewRWaxman https://orcid.org/0000-0002-7804-4025

References

AmericanChemistry Council (ACC) 2019a Email correspondenceACC2019bUSChemical Investment Linked to Shale Gas: $204

Billion andCounting (AmericanChemistry Council)(https://americanchemistry.com/Policy/Energy/Shale-Gas/Fact-Sheet-US-Chemical-Investment-Linked-to-Shale-Gas.pdf)

AllenDT 2016Emissions fromoil and gas operations in theUnitedStates and their air quality implications J. AirWasteManage.Assoc. 66 549–75

AllenDT et al 2013Measurements ofmethane emissions at naturalgas production sites in theUnited States Proc. Natl Acad. Soc.110 17768–73

AllenMR, Fuglestvedt J S, Shine KP, Reisinger A,Pierrehumbert RT and Forster PM2016Newuse of globalwarming potentials to compare cumulative and short-livedclimate pollutantsNat. Clim. Change 6 773–6

Alvarez RA, Pacala SW,Winebrake J J, ChameidesWL andHamburg S P 2012Greater focus needed onmethane leakagefromnatural gas infrastructure Proc. Natl Acad. Sci. 1096435–40

Alvarez RA et al 2018Assessment ofmethane emissions from theUS oil and gas supply chain Science 361 186–8 (http://sciencemag.org/lookup/doi/10.1126/science.aar7204)

Blum J 2019Texas rig count hits lowest point since January 2018Houston Chronicle (https://chron.com/business/energy/article/Texas-rig-count-hits-lowest-point-since-January-14028532.php)

BMI/Fitch 2018 BMIResearch—Fitch Solutions IndustryDatabase(https://fitchsolutions.com/bmi-research)

Brandt AR et al 2014Methane leaks fromNorthAmerican naturalgas systems Science 343 733–5

Brandt AR,MasnadiMS, Englander J G, Koomey J andGordonD2018Climate-wise choices in aworld of oil abundanceEnviron. Res. Lett. 13 044027

BurnhamA,Han J, ClarkCE,WangM,Dunn J B andPalou-Rivera I 2012 Life-cycle greenhouse gas emissions ofshale gas, natural gas, coal, and petroleum Environ. Sci.Technol. 46 619–27

Table 4.Total emissions inmillion tons of CO2e by Facility Type andProject Status.

Planned Construction Completed Unknown

Total

in TX

Total TX

and LA

%ofGrand

Total

Upstream 169.4 169.4 31.31%

Natural Gas Production 98.3 98.3 18.17%

Oil Production 71.1 71.1 13.14%

Midstream 101.3 9.3 10.8 0.0 40.4 121.4 22.4%

Compressor Station 11.0 0.0 0.4 11.0 2.04%

Gas Processing 4.6 0.4 4.4 5.0 0.92%

Pipeline, Gas 0.1 0.3 0.0 0.4 0.4 0.08%

Pipeline,NGL 0.4 0.0 0.0 0.0 0.5 0.5 0.09%

Pipeline,Oil 0.1 0.0 0.1 0.14 0.03%

Rail Terminal 0.0 0.0 0.0 0.005 0.00%

Storage, Gas 0.0 0.0 0.0 0.03 0.01%

Terminal, Oil 3.3 0.0 3.3 3.3 0.61%

Terminal, LNG 81.7 8.5 10.8 31.2 101.0 18.66%

Downstream 128.5 66.2 31.3 24.2 138.3 250.3 46.3%

Fractionator 9.2 1.9 0.0 11.1 11.1 2.06%

Petrochemical 113.9 36.9 31.3 24.2 107.3 206.3 38.1%

Refinery 5.4 27.5 0.0 19.9 32.9 6.07%

Total in TX (Mid- and

Downstream)114.5 41.9 16.9 5.4 178.7

GrandTotal 399.2 75.5 42.2 24.2 178.7 541.1

Note. Natural Gas andOil Production are restricted to vented, flared, or fugitive emissions from exploration, development, and production.

Petrochemical activities include awide variety of outputs, described by process in SI table S1.

10

Environ. Res. Lett. 15 (2020) 014004

CaiH,Han J, Elgowainy A andWangM2014 UpdatedVented,Flaring, and FugitiveGreenhouseGas Emissions for CrudeOil Production in theGREETModelReport (Lemont, IL:SystemsAssessmentGroup, Energy SystemsDivision,ArgonneNational Laboratory) (https://greet.es.anl.gov/files/emissions-crude-oil-2014)

Cathles LM, BrownL, TaamMandHunter A 2012A commentaryon ‘The greenhouse-gas footprint of natural gas in shaleformations’ byR.W.Howarth, R. Santoro, andAnthonyIngraffeaClim. Change 113 525–35

ChenH, Reilly J and Paltsev S 2019Did the ShaleGas BoomReduceUSCO2Emissions?Report (Cambridge,MA:MIT JointProgramon the Science and Policy of Global Change)

Davis S J, Caldeira K andMatthewsHD2010 Future CO2 emissionsand climate change from existing energy infrastructureScience 329 1330–3

DeLuchiMA1993 Emissions from the production, storage, andtransport of crude oil and gasolineAirWaste 431486–95

Dismukes D E, Terrell D andUptonGB 2019Gulf Coast EnergyOutlook (Baton Rouge, LA: E J Ourso College of BusinessEconomics & Policy Research Group and the Center forEnergy Studies, Louisiana State University)(https://lsu.edu/ces/publications/2018/gceo2019.pdf)

DuF andNojabaei B 2019A review of gas injection in shalereservoirs: enhanced oil/gas recovery approaches andgreenhouse gas controlEnergies 12 2355

EIA 2019Annual EnergyOutlook 2019with projections to 2050(Washington, DC: Energy InformationAgency) (https://eia.gov/outlooks/aeo/)

EIA 2018TheUnited States is now the largest global crude oilproducerToday in Energy (https://eia.gov/todayinenergy/detail.php?id=37053)

EIP 2018 31New or Expanded Petrochemical Plants Approved inHurricane Zone Along TX and LAGulf Coast EnvironmentalIntegrity Project(http://environmentalintegrity.org/news/31-new-or-expanded-petrochemical-plants/)

EIP 2019TrackingOil andGas Infrastructure EmissionsEnvironmental Integrity (http://environmentalintegrity.org/oil-gas-infrastructure-emissions/)

El-HoujeiriHM,MasnadiMS,VafiK,Duffy J andBrandtAR2018Oil ProductionGreenhouseGas Emissions EstimatorOPGEEv2.0 (http://pangea.stanford.edu/departments/ere/dropbox/EAO/OPGEE/OPGEE_documentation_v2.0b.pdf)

ElvidgeCD, BazilianMD, ZhizhinM,Ghosh T, BaughK andHsu F-C 2018The potential role of natural gasflaring inmeeting greenhouse gasmitigation targets Energy StrategyRev. 20 156–62

EPA 2017a Inventory of Greenhouse Gas Emissions& Sinks 1990-2015(Washington, DC:US Environmental ProtectionAgency,GreenhouseGas Reporting Program)

EPA2017b Inventory of Greenhouse Gas Emissions &Sinks 1990-2015Annexes (Washington,DC:USEnvironmentalProtectionAgency, GreenhouseGas Reporting Program)

EPA2019a Inventory ofUSGreenhouseGas Emissions and Sinks:1990–2017 (https://epa.gov/ghgemissions/inventory-us-greenhouse-gas-emissions-and-sinks-1990-2017)

EPA2019bUS Environmental ProtectionAgency Facility LevelGHGEmissionsData (http://ghgdata.epa.gov/ghgp/main.do)

Feyrer J,Mansur ET and Sacerdote B 2017Geographic dispersion ofeconomic shocks: evidence from the fracking revolutionAm.Econ. Rev. 107 1313–34

Fitch 2018 Fitch SolutionsUnited States Petrochemicals Report(https://store.fitchsolutions.com/united-states-petrochemicals-report)

Frischknecht R et al 2005The ecoinvent database: overview andmethodological framework Int. J. Life Cycle Assess. 10 3–9

Gaswirth S B, FrenchKL, Pitman J K,Marra KR,Mercier T J,Leathers-MillerHM, SchenkC J, TennysonME,Woodall CA, BrownfieldME, FinnTMandLe PA 2018Assessment of undiscovered continuous oil and gas resources

in theWolfcamp Shale and Bone Spring Formation of theDelaware Basin, PermianBasin Province, NewMexico andTexas, 2018 (Reston, VA) (http://pubs.er.usgs.gov/publication/fs20183073)

Graham J, Irving J, TangX, Sellers S, Crisp J,HorwitzD,Muehlenbachs L, Krupnick A andCareyD2015 Increasedtraffic accident rates associatedwith shale gas drilling inPennsylvaniaAccident Anal. Prevention 74 203–9

Hamit-HaggarM2012Greenhouse gas emissions, energyconsumption and economic growth: a panel cointegrationanalysis fromCanadian industrial sector perspective EnergyEcon. 34 358–64

HausmanC andKellogg R 2015Welfare and distributionalimplications of shale gasBrookings Papers Econ. ActivitySpring 2015 71–125

HayA2019US ethylene prices under pressure as supply builds, costslow ICIS (https://icis.com/explore/resources/news/2019/07/30/10398249/us-ethylene-prices-under-pressure-as-supply-builds-costs-low)

Hill E L andMaL 2017Does shale gas development impact infanthealth through drinkingwater?Working Paper p 49

Howarth RW2014Abridge to nowhere:methane emissions andthe greenhouse gas footprint of natural gas Energy Sci. Eng. 247–60

Howarth RW, Santoro R and Ingraffea A 2011Methane and thegreenhouse-gas footprint of natural gas from shaleformations: a letterClim. Change 106 679–90

InteragencyWorkingGroup 2016Technical Support Document:Technical Update of the Social Cost of Carbon for RegulatoryImpact Analysis Under ExecutiveOrder 12866 (InteragencyWorkingGroup on Social Cost of GreenhouseGases, UnitedStatesGovernment) (https://epa.gov/sites/production/files/2016-12/documents/sc_co2_tsd_august_2016.pdf)

IPCC2014Climate Change 2013—The Physical Science Basis byIntergovernmental Panel on Climate Change Cambridge Core(Cambridge: CambridgeUniversity Press) (https://doi.org/10.1017/CBO9781107415324)

Jenner S and LamadridA J 2013 Shale gas vs. coal: Policyimplications from environmental impact comparisons ofshale gas, conventional gas, and coal on air, water, and land intheUnited StatesEnergy Policy 53 442–53

JohnsenR, LaRiviere J andWolffH 2019 Fracking, coal, and airquality J. Assoc. Environ. Resour. Economists 61001–37

Kilian L 2016The Impact of the Fracking Boom onArabOil Producers(Rochester, NY: Social Science ResearchNetwork) (https://papers.ssrn.com/abstract=2733080)

Komarek T andCsehA 2017 Fracking and public health: Evidencefromgonorrhea incidence in theMarcellus Shale regionJ. Public Health Policy 38 464–81

MasnadiMS et al 2018Global carbon intensity of crude oilproduction Science 361 851–3

Newell RG andRaimiD 2014 Implications of shale gasdevelopment for climate changeEnviron. Sci. Technol. 488360–8

OGJ 2019 Permian gasflaring, venting reaches record highOil&Gas Journal (https://ogj.com/general-interest/hse/article/17279037/permian-gas-flaring-venting-reaches-record-high)

Olmstead SM,Muehlenbachs LA, Shih J-S, ChuZ andKrupnick A J2013 Shale gas development impacts on surfacewater qualityin Pennsylvania Proc. Natl Acad. Sci. 110 4962–7

PicardD1999 Fugitive Emissions fromOil andNatural Gas Activities(IPCC) (https://ipcc-nggip.iges.or.jp/public/gp/bgp/2_6_Fugitive_Emissions_from_Oil_and_Natural_Gas.pdf)

Ryan S 2012The costs of environmental regulation in aconcentrated industry Econometrica 80 1019–61

Subramanian R et al 2015Methane emissions fromnatural gascompressor stations in the transmission and storage sector:measurements andComparisons with the EPAGreenhouseGas Reporting ProgramProtocol Environ. Sci. Technol. 493252–61

11

Environ. Res. Lett. 15 (2020) 014004

TanakaK, CavalettO, CollinsW J andCherubini F 2019Assertingthe climate benefits of the coal-to-gas shift across temporaland spatial scalesNat. Clim. Change 9 389–96

TongD, ZhangQ, Zheng Y, Caldeira K, Shearer C,HongC,QinY andDavis S J 2019Committed emissions from existingenergy infrastructure jeopardize 1.5 °C climate targetNature572 373–7

UnruhGC2000Understanding carbon lock-inEnergyPolicy28817–30

WallaceC2019ThePermianBasin IsBoomingWithOil. But atWhatCost toWestTexans?TexasMonthly (https://texasmonthly.com/articles/permian-basin-booming-oil-cost-west-texans/)

Wigley TML2011Coal to gas: the influence ofmethane leakageClim. Change 108 601–8

ZimmerleD J et al 2015Methane Emissions from theNatural GasTransmission and Storage System in theUnited StatesEnviron. Sci Technol. 49 9374–83

12

Environ. Res. Lett. 15 (2020) 014004