Skip to main content
NCBI home page
ACS Omega logoLink to ACS Omega
. 2023 May 19;8(22):19443–19454. doi: 10.1021/acsomega.3c00676

Chemical Characterization of Natural Gas Leaking from Abandoned Oil and Gas Wells in Western Pennsylvania

Dominic C DiGiulio †,*, Robert J Rossi ‡, Eric D Lebel ‡, Kelsey R Bilsback ‡, Drew R Michanowicz ‡, Seth BC Shonkoff ‡,§,∥
PMCID: PMC10249135  PMID: 37305312

Abstract

graphic file with name ao3c00676_0006.jpg

The U.S. Environmental Protection Agency estimates that there are over 3.2 million abandoned wells in the United States. Studies conducted on gas emissions from abandoned wells have been limited to methane, a powerful greenhouse gas, due to concerns regarding climate change. However, volatile organic compounds (VOCs), including benzene, a known human carcinogen, are known to be associated with upstream oil and gas development and hence could also be released when methane is emitted to the atmosphere. In this investigation, we analyze gas from 48 abandoned wells in western Pennsylvania for fixed gases, light hydrocarbons, and VOCs and estimate associated emission rates. We demonstrate that (1) gas from abandoned wells contains VOCs, including benzene; (2) VOCs are emitted from abandoned wells, the magnitude of which depends on the flow rate and concentration of VOCs in the gas stream; and (3) nearly one-quarter of abandoned wells are located within 100 m of buildings, including residences, in Pennsylvania. Together, these observations indicate that further investigation is necessary to determine whether emissions from abandoned wells pose an inhalation risk to people living, working, or congregating near abandoned wells.

1. Introduction

The U.S. Environmental Protection Agency (EPA) estimates that there are over 3.2 million abandoned wells in the United States.1 This number includes oil and gas wells with no recent production (plugged, inactive, temporarily abandoned, shut-in, and idle), with or without a responsible owner (orphaned). An on-line review of records from the Pennsylvania Department of Environmental Protection (PADEP)2 indicates that there are 24,619 documented abandoned wells in Pennsylvania, of which 18,608 have associated geographical coordinates (Figure 1). However, the PADEP estimates that there are approximately 200,000 abandoned oil and gas wells that remain unaccounted for in state records.3,4

Figure 1.

Figure 1

Open in a new tab

Locations of documented abandoned wells in western Pennsylvania having coordinates,2 plugged and unplugged abandoned wells screened but not sampled in this investigation, and plugged and unplugged abandoned wells both screened and sampled in this investigation. State park5 and state forest6 boundaries are included where screening and sampling occurred. The polygon in southwestern Pennsylvania represents the outline of Allegheny County.

Leakage of natural gas to the atmosphere at the wellhead at active and abandoned wells can occur through poorly sealed joints, flanges, and valves. Leakage can also occur through a casing vent engineered to allow gas flow between the annular spaces of the surface, intermediate, and production casing, collectively commonly referred to as surface casing vent flow (SCVF).7 Venting effectively eliminates pressure buildup in an annular space, which can cause the entry of gas into groundwater.8 Casing vents are commonly used in Canada, especially Alberta,9 British Columbia,10 and Quebec.11 However, the use of casing vents in the United States appears limited to Pennsylvania.7 In Pennsylvania, plugged oil and gas wells penetrating “marketable” coal must be equipped with a 5 cm (2 in.) vent pipe 9.1 m (30 ft) below the coal bed.12

When subsurface gas leakage occurs along or away from the borehole, it is often referred to as gas migration.13 Gas migration can result in the efflux of gas through soil to the surface near oil and gas wells or transport to water wells14 or surface water15,16 through deeper and more extensive lateral migration in groundwater. Between 1987 and 2013, there were 133 confirmed cases of gas migration to water wells in Pennsylvania attributed to abandoned wells.3 Gas migration from abandoned wells into buildings and residences has resulted in emergency plugging actions17,18 and contributed to explosions in Pennsylvania19 and elsewhere.20 Gas migration and SCVF are believed to be due to a lack of cement in non-producing but gas-containing strata overlying reservoirs, defects in cement during setting, poor bonding between cement and casing or cement and boreholes, drilling induced fractures outside boreholes,21,22 and corrosion of casing, especially in the presence of gas containing hydrogen sulfide (H2S).20

Concern regarding climate change has led to considerable interest in quantifying emissions rates (mass per unit time) of methane, a powerful greenhouse gas, from active23,24 and abandoned oil and gas wells.25−36 To date, there is only one published study where the measurement of volatile organic compounds (VOCs) in gas from an abandoned well was attempted.29 However, after dilution of gas from the well with air inside an enclosure covering the well, benzene was not detected at the reporting limit of 4.2 parts per billion volume (ppbv).

In addition to fixed gases, which include oxygen (O2), argon (Ar), nitrogen (N2), carbon dioxide (CO2), and hydrogen (H2), and light hydrocarbons, which include methane, ethane, propane, isobutane, n-butane, isopentane, and n-pentane, natural gas is known to contain H2S25 and VOCs such as benzene, toluene, ethylbenzene, and xylene (BTEX components).37−40 Benzene, a known human carcinogen,41 is of particular concern when abandoned wells are in close proximity to buildings, including residences. The EPA estimates that continuously breathing air containing 1.3–4.5 μg m–3 (equivalent to 0.40–1.4 ppbv at 293 K and 1 atm) would result in not greater than a one-in-a hundred thousand increased chance of developing cancer.41 The California chronic reference exposure level (REL) for benzene is 3 μg m–3 (∼0.94 ppbv at 293 K and 1 atm).42

The objectives of this investigation are to determine the (1) composition (fixed gases, light hydrocarbons, VOCs, and H2S) of gas emitted from wellheads or open casings of abandoned wells; (2) emission rates of light hydrocarbons and VOCs from these wells; and (3) proximity of abandoned wells to buildings, including residences, in western Pennsylvania. Taken together, the presence of these conditions would indicate the need for further investigation to determine whether emissions from abandoned wells pose an inhalation risk to people living, working, or congregating near abandoned wells.

2. Materials and Methods

2.1. Abandoned Well Screening

Between May and August 2021, the PADEP screened 90 abandoned wells (29 plugged, 61 unplugged) for natural gas leakage in western Pennsylvania (Figure 1). Abandoned wells were screened by placing the probe extension of a Bascom-Turner Explorer EGA 612 portable gas analyzer directly over the open casing, joints, fittings, and valves of unplugged wells and the riser vent pipe of plugged wells (Figure S1). The instrument is equipped with a dual catalytic combustion and thermal conductivity detector for measurement response readings between 20 parts per million volume (ppmv) and 100% gas.43 The EGA 612 was calibrated daily to 1.0% methane. The instrument is also equipped with electrochemical cells for the detection of H2S and carbon monoxide (CO) at a lower limit of detection of 1 ppmv.43 Additional information on the EGA 612 is provided in Section S1.

To measure gas flow from abandoned wells and instrument response in the absence of interference from ambient air, following initial screening at abandoned wells exhibiting leakage, the PADEP enclosed primary points of leakage and repeated measurements with the EGA 612. Depending on the configuration of leakage, enclosure materials included 3.8–363 L (1–96 gallon) polyethylene bags, ziploc sandwich bags, flexible couplings and adapters, poly(vinyl chloride) reducer fittings, and 0.95 cm (3/8″) brass fittings. A stainless-steel Marlin spike was used to create a hole in polyethylene and ziploc bags for the insertion of 0.95 cm internal diameter polyethylene tubing (Figure S2). With the exception of a plugged well (059-01162), where an orifice meter was used because of high flow (146 m3 d–1), polyethylene tubing was connected to a low-flow Alicat Whisper flowmeter (details on operating principles provided in Section S2) for flow measurement. Since readings often fluctuated during measurement, a maximum value was recorded after relative stabilization (minimum oscillation). Fluctuation during flow measurement resulted in an unknown level of error. The flowmeter was factory calibrated with methane prior to field use, with minimum and maximum flow detections of 20 mL min–1 and 9.4 L min–1, respectively.

A polycarbonate cone (to minimize the effect of wind) attached at the end of the EGA 612 sample hose was then used to screen gas concentration directly above soil (Figure S3) within 1 m of abandoned wells in at least three locations around each well. The cone was held in place for at least 3 s prior to advancement. At some locations, the presence of standing water prevented soil screening. The results of all screening activities between May and August 2021 to support this investigation were provided by the PADEP.

2.2. Abandoned Well Sampling and Analysis

Between September and November 2021, with assistance from the PADEP, we screened (using the same procedures previously described) and sampled gas from 8 plugged abandoned wells and 40 unplugged abandoned wells for fixed gases, light hydrocarbons, and VOCs. At one well (053-25266), we collected two samples because of the detection of leakage from two open pipes—one from the production casing (053-25266p) and one from the annular space outside production casing (053-25266a). Seven active wells were sampled but not screened for leakage of gas (as per the access agreement). Flow measurements using the Alicat Whisper flowmeter were corrected based on the viscosities and molecular weights of fixed gases and light hydrocarbons detected in gas mixtures (Section S2). After flow correction, the lower range of detectable flow varied from 12 to 17 mL min–1.

Well names, API numbers, geographic coordinates, and information on completion dates, target depths, and formations (where known) of sampled abandoned wells are provided in Table S1. Sampled abandoned unplugged well locations included 7 wells in Oil Creek State Park (OCSP), 11 wells in Hillman State Park (HSP), and 7 wells in Cornplanter State Park (CPSP) (Figure 1). Sample locations were selected based on public accessibility and locations where the PADEP could obtain private access to abandoned wells. A brief background on oil and gas development at OCSP, HSP, and CPSP is provided in Supporting Information (Section S3).

After field screening and flow measurement, we used scissors to cut the base of 18 × 18 cm (7″ × 7″), 30 × 30 cm (12″ × 12″), or 47 × 47 cm (18.5″ × 18.5″) 5.0 mil perfluoroalkoxy (PFA) or 2.5 mil fluorinated ethylene propylene (FEP) gas sampling bags equipped with an open Kynar valve to secure an area of leakage using rigid plastic zip ties. We utilized the smallest PFA or FEP bag size as practical to enclose leakage locations to minimize the surface area for sealing, heat transfer (condensation) during enclosure, and time to reach equilibrium, as recommended in the EPA’s Protocol for Equipment Leak Estimates.44 Condensation on the inside of sampling bags was absent during sampling.

We then connected a short section (30–60 cm) of 0.48 × 0.64 cm (3/16″ × 1/4″) polytetrafluoroethylene (PTFE) tubing to the Kynar valve and connected the tubing to another 30 × 30 cm PFA or FEP sample bag acting as a plenum (Figure S4). This sampling procedure was pursued to enable passive sample collection (no active gas extraction from the enclosure) to avoid, to the extent possible, the intrusion of atmospheric air into the plenum. We used PFA and FEP sample bags and PTFE tubing to minimize off-gassing or adsorption of VOCs commonly associated with Tedlar bags and polyethylene tubing.45,46 When necessary, we used electrical tape completely on the outside of sample bags to assist in securing sampling bags.

Depending on conditions at abandoned wells, we allowed plenum bags to inflate between 0.5 and 42 h (Table S2) prior to extraction of gas from the plenum bags into one-liter evacuated summa canisters using 67, 200, and 1000 mL min–1 single-use stainless-steel flow regulators. Summa canisters and regulators were provided by Eurofins Air Toxics Laboratory, Folsom, California. We collected duplicate samples at one plugged abandoned well and at two active wells using single-use 0.64 cm outside diameter stainless-steel tees. We also collected one replicate sample at an abandoned unplugged well by sampling the well twice within a ten-minute period. Variation in concentration was calculated as the relative percent difference (RPD), defined as 2*100|R1-R2|/(R1+R2), where R1 = sample 1 and R2 = sample 2. RPDs of fixed gases and light hydrocarbons were below 10%. The RPDs for VOCs were below 20%.

We collected equipment blanks in three 30 × 30 cm FEP bags, two 30 × 30 cm PFA bags, and one 47 × 47 cm PFA bag by inflating sample bags with ultra-pure N2 gas. We sealed the sample bags using PTFE tubing and 1000 mL min–1 dedicated flow controllers and allowed the sample bags to equilibrate for a 24 h period. Samples were subsequently extracted into one-liter evacuated summa canisters. At one abandoned well (125-21260), sample collection necessitated the use of duct and electrical tape overlapping cut sample bags and metal from the well, thereby providing a potential pathway for sampled air to contact adhesive. To evaluate the potential effect of off-gassing from electrical or duct tape at this well, we cut two 18 × 18 cm FEP bags at the base and sealed one using electrical tape and the other using duct tape. We handled these equipment blanks in the same manner as other equipment blanks. We also collected two samples of atmospheric air distant (e.g., >1 km) from abandoned wells (one at OCSP and one at HSP) in six-liter evacuated summa canisters to evaluate the presence of VOCs in the background atmospheric air.

We shipped summa canisters to the Eurofins Air Toxics Laboratory for analysis of fixed gases and light hydrocarbons using Modified ASTM D-1945 (gas chromatography/flame ionization detector or thermal conductivity detector) and for analysis of VOCs plus naphthalene using the EPA Method TO-15 (gas chromatography/mass spectrometry) in the full scan mode using the same canister. Analysis of fixed gases included N2, O2 + Ar, (on the analytical column used for analysis, O2 coeluted with Ar), CO2, and H2. Analysis of light hydrocarbons included methane, ethane, propane, isobutane, n-butane, neopentane, isopentane, n-pentane, and hexane-plus. VOCs (as defined in EPA Method TO-15) analyzed are summarized in Table S3. Compound reporting limits were determined by the gas dilution necessary to quantitate compounds detected at high concentrations. Eurofins is accredited by the National Environmental Laboratory Accreditation Program. Eurofins provided full documentation on quality assurance procedures and results associated with analysis.

To determine the stable isotope ratio of carbon in methane (δ13C–CH4) relative to the Vienna Peedee belemnite standard, we extracted 200–400 mL of gas from a plenum or secured sample bag using a bulb syringe into a 1 L 5 mil (0.14 mm) thick multi-layer foil Cali-5 Bond bag for transportation and storage.

We analyzed samples using a Picarro Cavity Ring Down Spectrometer (model G2210-i) within 2–5 weeks of sample collection with an accuracy of ±0.5‰. We periodically checked the analyzer with isotopic standards to ensure that the results were within 10% of the certified values. Additional information on stable carbon isotope analysis is provided in Section S4.

2.3. Estimation of Intrusion of Air into Sample Bags

During initial sampling efforts, it became apparent that 5.0 mil PFA sample bags were too thick and pressure at leakage locations on abandoned wells was too low to allow inflation of most plenum bags. In some instances, this necessitated sample collection directly from secured PFA bags. In addition, the rigidity of PFA bags made it difficult to obtain a good seal around a point of leakage. While we achieved greater success transitioning to 2.5 mil FEP sample bags, in some cases, flow from abandoned wells was still too low to allow inflation of FEP plenum bags even at inflation times exceeding 40 h. This necessitated sample collection directly from secured FEP sample bags. Information on flow rates during sampling, enclosure times, and flow rates of controllers during sampling is summarized in Table S2.

Intrusion of atmospheric air into a sampling vessel will result in elevated O2 and N2 concentrations and, without adjustment, will result in biased low estimates of light hydrocarbons and VOCs. To estimate concentrations of fixed gases, light hydrocarbons, and VOCs in samples, we used ratios of N2, O2 + Ar, and CO2 in atmospheric air to adjust sample results when air intrusion was plausible (N2 levels above detection). We calculated upper and lower bounds of fixed gas, light hydrocarbon, and VOC concentrations by assuming concentrations of O2 + Ar in samples at 0.0 and 3.0%, respectively, based on available information from oil and gas wells in Devonian-age reservoirs in Pennsylvania37,47−49 (Figure S5) and three abandoned wells at HSP having non-detectable levels of N2 (<0.19%) and levels of O2 + Ar between 2.4 and 2.9% (indicating that O2 + Ar levels in samples this high were not due to air intrusion). When N2 levels were above detection limits, we used the midpoint of the upper and lower bounds to report concentrations. A detailed discussion of concentration adjustment and equation development is provided in Section S5.

2.4. Statistical Analysis

Environmental datasets containing sample results are often left-censored (sample results below reporting limits). Substitution methods such as assigning values of zero, one-half of a reporting limit, or a reporting limit itself to sample results below a reporting limit are commonly used to calculate the mean and standard deviation of left-censored data.50 However, substitution methods have no statistical basis,51 perform poorly in datasets with multiple reporting limits,51 and commonly result in overestimation or underestimation of a mean and standard deviation.52 The EPA now discourages the use of substitution methods in calculating the mean, standard deviation, and other statistical parameters (e.g., 95% upper confidence limit) of left-censored data.53 Several methods have been used for calculating statistical parameters of left-censored data, including maximum likelihood estimation, regression on order statistics,51 and the non-parametric Kaplan–Meier (KM) method.54 The KM method used here appears to provide the most accurate estimate of statistical parameters containing up to 70% left-censored data.55 Hypothesis testing was conducted using the non-parametric Tarone–Ware test,56 which also accounts for left-censored data. The EPA’s ProUCL (version 5.2) software package53 was used to calculate the KM mean, KM standard deviations, and conduct hypothesis testing. The calculation of median levels was handled by ranking reporting limits. When more than one-half of a dataset contained left-censored data, a median was reported as less than the reporting limit.

2.5. Proximity Analysis

During our sampling campaign, we observed several abandoned wells located within 100 m of residential buildings. To better understand how commonly this occurs, we determined the distance from the nearest building to the location of abandoned wells. The locations of abandoned wells were taken from the PADEP orphaned and abandoned wells report.2 Building locations were taken from building footprint polygons generated by Microsoft using deep learning algorithms.57 The distance of each well from any building was determined using the sf package in R.58 We limited a proximity analysis of residential buildings to one county (Allegheny County) because of the ease of accessibility of parcel tax assessment data.59 Parcels having residents were identified by first excluding any assessment with descriptor codes of “utilities” or “industrial” in R.60 We then selected every assessment with a descriptor code of “residential” and appended these with use codes containing keywords, such as “single family” or “condominium”. After identifying the parcels containing residents, buildings contained within these parcels61 were identified via spatial joining in the sf package in R.58 To account for edge effects, a 2.5 km buffer was extended from the edge of Allegheny County, and the proximity of wells contained within this buffer to Allegheny County residential buildings was calculated.

To assess the accuracy of the proximity analysis, we randomly selected 5% (n = 213) of abandoned wells that were found to be within 100 m of a building and cross-referenced them with the building footprint dataset with either 0.6 m (Surdex Corporation)62 or 15.4 cm (Quantum Spatial, Inc.)63 resolution aerial imagery captured in 2019 and 2018, respectively. In some cases, we also used historic imagery in Google Earth Pro64 for cross-referencing. For each well, we classified the observation as a “true positive” (i.e., a structure in which people would reasonably be inside of), a “false positive” (e.g., ruins of a former building, oil and gas development infrastructure), and in one case “demolished”. Additional detail is provided in Section S6.

2.6. Comparison with Published Sources of VOCs from Active Oil and Gas Wells

Finally, we conducted a database search to compare the sample results of VOCs from gas collected from active and abandoned wells in this investigation with those from active wells at other locations. The Colorado Oil and Gas Conservation Commission maintains an extensive electronic database38 of bradenhead (annular space between surface and production casing) (n = 115) and production gas (PG) samples (n = 251) which contain data on VOCs including benzene. Search terms for PG included production, production casing, production string, oil/gas wellhead, intermediate casing, separator, gas stream, and gas sales point. We also extracted data from a database37 maintained by the United States Geological Survey with PG samples (n = 23) containing data on benzene concentration. Published sources of information on VOCs in natural gas include production and bradenhead (n = 25 each) samples from the Pavillion, WY Field,39 and a report40 to the Texas Commission on Environmental Quality summarizing results of gas samples (n = 117) in the Anadarko, Ben Arch-Fort Worth, East Texas, Permian, and Western Gulf Basins prior to glycol dehydration.

3. Results and Discussion

3.1. Field Screening

During open screening, the frequency of EGA 612 instrument response above detection (20 ppmv) at plugged wells (9/27 ∼ 33%) was less than that of unplugged wells (53/60 ∼ 88%) (p < 0.001) (only enclosed screening was conducted at 3 abandoned wells). This observation is consistent with the findings of others10,27,34 that plugging decreases the incidence of leak detection at the wellhead of abandoned wells. Gas leakage at unplugged wells was due to the presence of ground-level open casing when wellheads were absent or open pipe attached to wellheads or corroded casing when wellheads were present. When detected, the mean concentrations of gas during open screening were 26 and 14% at plugged and unplugged wells, respectively, while the mean concentrations of gas screened with enclosure were 38 and 37% at plugged and unplugged wells, respectively. Hence, detected gas concentrations from leakage locations at abandoned wells were generally high (often above the lower explosive limit of methane of 5%) and increased, as expected, with enclosure using sample bags (Figure S6) due to a lack of dilution from atmospheric air.

During field screening, H2S was detected at 3 of 90 (∼3.3%) abandoned wells. H2S concentrations were 1.5 and 5.0 ppmv at leakage locations of two unplugged wells and 11.2 ppmv from the vent of plugged well (059-01162), which, because of a high flow rate, had an H2S emission rate of 2.3 g d–1. The gas from this well consisted of 92% methane and 6.6% N2 with no detection of VOCs. El Hachem and Kang25 detected H2S emissions at 3 of 20 (∼15%) abandoned wells in Ontario, Canada, with a maximum emission rate of 144 g d–1. Given the toxicity of H2S (California acute and chronic inhalation REL at 42 and 10 μg m–3, respectively, or 29 and 7 ppbv, respectively, at 293 K and 1 atm),42 H2S should be routinely screened at abandoned wells, especially in areas where gas from oil and gas wells are known to have high levels of H2S. We also detected CO at 17 of 90 (∼19%) abandoned wells at a maximum concentration of 290 ppmv and a maximum emission rate of 5.7 g d–1. The California acute REL for CO is 20 ppmv.42 This finding indicates that CO should also be routinely screened at abandoned wells.

Gas was detected at the ground surface at 7 of 61 (∼11%) unplugged wells and 3 of 29 (∼10%) plugged wells within a 1 m radius of the wells. At one plugged well, ground-level gas was measured at a concentration of 6% with no detection at the vent pipe of the well. At another plugged well, ground-level gas was measured at a concentration of 0.05%. There was no vegetation within 2 m of this well, suggesting high root zone gas concentrations, potentially from displacement of oxygen. These observations highlight the need to evaluate gas flux at the surface before and after plugging.

Soil screening, as performed here, has been utilized by others65 to detect natural gas migration to the surface. However, soil screening is generally recognized as having insufficient sensitivity to effectively detect gas efflux from soil to the atmosphere.66 Soil-gas sampling and flux chambers are more sensitive methods of detection and measurement of gas migration in soil near and distant from abandoned oil and gas wells67−71 especially when methane-oxidizing microorganisms reduce methane efflux,72−74 and efflux is affected by variation in barometric pressure.68 Hence, the frequency of detection of gas emissions at the soil surface (∼10%) from abandoned wells observed in this investigation could be an underestimate to some unknown degree.

3.2. Composition of Gas from Active and Abandoned Wells

Gas from active and abandoned wells sampled in this investigation was from Devonian-age reservoirs (Table S1), as is the case for conventional oil and gas production throughout the Appalachian Basin.75 At the abandoned well where two samples were collected, the gas sample from the annular space (053-25266a) indicated a greater proportion of microbial to thermogenic gas (δ13C–CH4 = −60.3‰) compared to the gas sample from production (053-25266p) (δ13C–CH4 = −54.3‰) (Figure S7). With the exception of gas sample 053-25266a, δ13C–CH4 values and ratios of methane to ethane and propane were within the range of expected values for thermogenic gas originating from Devonian-age reservoirs.47−49 These observations suggest that gas from abandoned wells originated from Devonian-age reservoirs or that gas in overlying strata potentially in contact with corroded or compromised casing had stable isotopic values for methane similar to those of Devonian-age reservoirs.

Analyses of gas samples from equipment blanks indicated no impact on sample results (Section S7 and Figure S8). Despite observations of atmospheric intrusion into enclosures during sampling at some locations (Figure S9), the estimation technique utilized here effectively constrained lower and upper estimates of fixed gas, light hydrocarbon, and VOC concentrations (Tables S4 and S5, lower and upper estimates of benzene illustrated in Figure S10). For instance, the maximum difference in the lower and upper bound estimates for benzene was only 32 ppmv (lower and upper bound estimates of 196 and 228 ppmv, respectively, at well 125-21260).

Summary statistics for fixed gases, light hydrocarbons, and VOCs in gas samples from active and abandoned wells in western Pennsylvania are provided in Table 1. Differences in concentrations of fixed gases, light hydrocarbons, and VOCs in active and abandoned wells (Figure S11) were not statistically significant (p > 0.05). Hexane, cyclohexane, and heptane were frequently detected (4.1–8.1% left-censored data sets) in gas samples from abandoned wells (Table 1) and dominated VOC analysis (Figures S12 and S13). The maximum concentration of hexane (14,000 ppmv) was greater than the California chronic REL42 (7000 μg m–3 or 1.99 ppmv at 293 K and 1 atm) by a factor of 7000.

Table 1. Summary Statistics for the Levels of Fixed Gases, Light Hydrocarbons, and VOCs in Gas Samples from Active and Abandoned Wells in Western Pennsylvaniaa.

  active wells
 abandoned wells
compound frequency of left-censored data minimum maximum KM mean KM standard deviation median frequency of left-censored data minimum maximum KM mean KM standard deviation median
Fixed Gases as Measured using ASTM-D-1945 (%)
oxygen + argon 4/7 <0.17 1.8 0.45 0.56 <0.18 5/49 <0.19 2.9 1.4b 0.60b 1.5b
nitrogen 6/7 <0.17 4.3     <0.19 16/49 <0.18 20 7.9 6.9 8.9
carbon dioxide 0/7 0.024 0.084 0.044 0.020 0.040 4/49 <0.18 26 2.9 5.5 0.43
hydrogen 0/7 0.023 0.27 0.13 0.086 0.15 37/49 <0.018 7.2 0.26 1.1 <0.056
Natural Gas constituents as Measured using ASTM-D-1945 (%)
methane 0/7 28 97 83 83 91 0/49 17 98 78 78 83
ethane 0/7 2.7 29 8.2 9.3 5.5 2/49 <0.039 20 3.8 4.7 2.2
propane 0/7 0.18 26 4.8 9.4 1.5 4/49 <0.0019 25 2.5 4.6 0.42
isobutane 0/7 0.017 3.7 0.73 1.3 0.22 4/49 <0.0019 4.6 0.61 1.1 0.10
butane 0/7 0.025 8.5 1.5 3.1 0.33 5/49 <0.0019 10 1.2 2.2 0.13
neopentane 0/7 0.0045 0.011 0.012 0.012 0.0087 31/49 <0.0018 0.10 0.0092 0.017 <0.0064
isopentane 0/7 0.0087 1.9 0.37 0.68 0.15 7/49 <0.0019 5.5 0.54 1.0 0.058
pentane 0/7 0.002 1.5 0.29 0.54 0.12 13/49 <0.0018 4.7 0.43 0.83 0.042
hexane+ 1/7 <0.018 1.5 0.34 0.48 0.15 18/49 <0.018 6.2 0.67 1.2 0.13
VOCs as Measured using the EPA Method TO-15 (ppmv)
hexane 0/7 1.8 3900 790 1400 310 2/49 <0.00094 14000 1400 2400 160
cyclohexane 0/7 0.38 780 180 270 79 2/49 <0.00094 2700 340 533 56
heptane 0/7 0.32 1000 232 350 94 6/49 <0.00094 2400 400 570 79
benzene 1/7 <0.22 170 33 56 12 14/49 <0.00094 250 36 65 2.8
toluene 1/7 <0.22 180 36 59 16 29/49 <0.00094 250 21 50 <2.1
ethylbenzene 7/7 <0.22 <22     <2.0 41/49 <0.00094 40 2.7 7.7 <1.3
m,p-Xylene 2/7 <0.22 58 12 19 5.3 28/49 <0.00094 340 22 57 <2.1
o-Xylene 5/7 <0.22 3.2     <3.2 37/49 <0.00094 55 4.2 10 <1.7
1,3,5-trimethylbenzene 7/7 <0.22 <22     <2.0 39/49 <0.00094 30 1.8 4.9 <1.7
1,2,4-trimethylbenzene 7/7 <0.22 <22     <2.0 40/49 <0.00094 40 1.6 6.2 <1.4
4-ethyltoluene 7/7 <0.22 <22     <2.0 40/49 <0.00094 30 1.6 4.8 <1.4
propylbenzene 7/7 <0.22 <22     <2.0 46/49 <0.00094 5.7 0.26 0.99 <1.3
cumene 7/7 <0.22 <22     <2.0 45/49 <0.00094 5.5 0.25 0.93 <1.1
Open in a new tab
a

Midpoints of adjusted fixed gas, light hydrocarbon, and VOC concentration data were used for all calculations. KM mean and standard deviations were not calculated for data sets having less than 3 detected values. All data reported two significant figures.

b

Mean and median values of 1.5% for O2 + Ar in abandoned wells are an artifact of using 0.0 and 3.0% O2 + Ar to estimate the upper and lower bounds of air intrusion.

Benzene was also frequently detected in gas from abandoned wells (28.6% left-censored data set), with less detection frequency of toluene (59.2% left-censored data set) and m,p-xylene (57.1% left-censored data set) (Table 1). The maximum concentrations of benzene (250 ppmv), toluene (250 ppmv), and m,p-xylene (340 ppmv) were greater than the California chronic RELs42 by factors of 250,000, 2000, and 2000, respectively (REL for toluene = 420 μg m–3 or 0.111 ppmv and REL for xylenes = 700 μg m–3 or 0.161 ppmv at 298 K and 1 atm). Hence, while VOCs other than benzene were detected in gas from abandoned wells, observed benzene concentrations largely the dominated potential risk of exposure. The mean values of these compounds were substantially greater than the median values, indicating a highly skewed distribution of data with high concentrations controlling the calculation of the mean.

Ethylbenzene, 1,3,5-trimethylbenzene, 1,2,4-trimethylbenzene, 4-ethyltoluene, propylbenzene, and cumene (isopropylbenzene) were detected infrequently (75.5–93.9% left-censored data sets) in abandoned wells at maximum concentrations of 40, 30, 40, 30, 5.7, and 5.5 ppmv, respectively (Table 1). Summary statistics with left-censored data sets in excess of 70% are considered to be unreliable.53 Halogenated compounds, ketones (acetone, 2-butanone, 2-hexanone, 4-methyl-2-pentanone), alcohols (ethanol, propanol), styrene, naphthalene, and 1,3-butadiene were not detected in any samples. Other compounds (e.g., methyl cyclohexane and mixed isomers of butane and pentane) that have been detected in gas samples collected directly from oil and gas wells elsewhere38 were not included in the EPA Method TO-15 utilized for this investigation. With the exception of the detection of acetone and 2-propanol at maximum levels of 7.2 and 2.7 ppbv, respectively, VOCs were not detected in atmospheric air samples distant from abandoned wells. Hence, background air quality had no impact on the sample results.

The mean and median concentrations of benzene detected in gas from active and abandoned wells in this investigation were generally less than those observed in active wells elsewhere in the United States (Figure 2). This finding suggests that VOC concentrations in gas from abandoned wells in other areas of the United States could be higher than those measured in this investigation, necessitating further investigation elsewhere.

Figure 2.

Figure 2

Open in a new tab

Detected benzene concentrations reported in PG and bradenhead gas (annular space between production or intermediate casing and surface casing) at the Pavillion Field, WY;39 in Colorado (CO);38 in Bend Arch-Fort Worth, East Texas, Permian, and Western Gulf Basins40 in Texas (TX); additional locations in Texas, Louisiana (LA), and California (CA);37 and midpoints of lower and upper bounds of detected (above reporting limits) concentrations in this study. The mean and median of detected concentrations of benzene in abandoned wells in this study were 53 and 2.8 ppmv, respectively.

3.3. Emission Rates of Light Hydrocarbons and VOCs from Abandoned Wells

Summary statistics for emission rates of light hydrocarbons and VOCs from abandoned wells in western Pennsylvania are provided in Table 2. Methane dominated emissions of light hydrocarbons at mean values of 2.2 kg d–1—more than two orders of magnitude greater than the mean values of the next highest emitting light hydrocarbon, ethane, at 18 g d–1. The mean and median emission rates of light hydrocarbons generally decreased with increased carbon numbers (methane > ethane > propane > butane ≈ isobutane > isopentane ≈ pentane > neopentane).

Table 2. Summary Statistics of Emission Rates (g D–1) of Light Hydrocarbons and VOCs from Abandoned Wells in Western Pennsylvaniaa.

compound frequency of left-censored data minimum maximum KM mean KM standard deviation median
Natural Gas Constituents as Measured using ASTM-D-1945 (g d–1)
methane 15/48 <1.8 × 10+0 8.3 × 10+4 2.2 × 10+3 1.2 × 10+4 6.9 × 10+1
ethane 15/48 <4.2 × 10–3 3.8 × 10+2 1.8 × 10+1 5.7 × 10+1 1.5 × 10+0
propane 18/48 <5.3 × 10–3 6.8 × 10+1 3.9 × 10+0 1.3 × 10+1 4.2 × 10–1
isobutane 18/48 <1.8 × 10–3 1.3 × 10+1 7.5 × 10–1 2.4 × 10+1 1.1 × 10–1
butane 18/48 <8.5 × 10–4 2.3 × 10+1 1.2 × 10+0 3.9 × 10+1 1.5 × 10–1
neopentane 24/38 <3.5 × 10–4 7.3 × 10–1 3.0 × 10–2 1.1 × 10–1 <5.4 × 10–3
isopentane 18/48 <8.5 × 10–4 8.8 × 10+0 5.0 × 10–1 1.5 × 10+1 7.2 × 10–2
pentane 26/48 <7.3 × 10–4 7.1 × 10+0 3.5 × 10–1 1.2 × 10+1 5.7 × 10–2
VOCs as Measured using EPA Method TO-15 (g d–1)
hexane 17/48 <5.7 × 10–6 1.4 × 10+1 7.2 × 10–1 2.3 × 10+1 1.1 × 10–1
cyclohexane 17/48 <1.2 × 10–5 3.0 × 10+0 1.6 × 10–1 4.9 × 10–1 2.7 × 10–2
heptane 21/48 <5.9 × 10–6 5.7 × 10+0 3.8 × 10–1 1.1 × 10+1 5.2 × 10–2
benzene 24/48 <3.0 × 10–6 7.4 × 10–1 2.2 × 10–2 1.1 × 10–1 2.2 × 10–3
toluene 35/48 <3.5 × 10–6 8.7 × 10–1 2.7 × 10–2 1.3 × 10–1 <1.0 × 10–3
ethylbenzene 43/48 <4.0 × 10–6 4.4 × 10–2 1.8 × 10–3 8.1 × 10–3 <7.2 × 10–4
m,p-xylene 31/48 <4.0 × 10–6 4.3 × 10–1 2.1 × 10–2 8.3 × 10–2 <1.9 × 10–3
o-xylene 40/48 <4.0 × 10–6 8.4 × 10–2 3.9 × 10–3 1.6 × 10–2 <9.9 × 10–4
1,3,5-trimethylbenzene 40/48 <4.6 × 10–6 4.4 × 10–2 1.6 × 10–3 6.5 × 10–3 <1.1 × 10–3
1,2,4-trimethylbenzene 41/48 <4.6 × 10–6 5.8 × 10–2 1.7 × 10–3 8.4 × 10–3 <1.0 × 10–3
4-ethyltoluene 41/48 <4.6 × 10–6 4.2 × 10–2 1.4 × 10–3 6.3 × 10–3 <1.0 × 10–3
propylbenzene 46/48 <4.6 × 10–6 8.1 × 10–3     <7.7 × 10–4
cumene 45/48 <4.6 × 10–6 6.2 × 10–3 1.7 × 10–4 9.4 × 10–4 <7.7 × 10–4
Open in a new tab
a

The product of the lower limits of flow detection and reporting limits was used for the calculation of minimum emission rates. Midpoints of adjusted light hydrocarbon and VOC concentration data were used for all calculations. KM mean and standard deviations were not calculated for data sets having less than three reportable values. All data and calculations are reported in two significant figures.

We observed mean values of methane emission rates for plugged and unplugged abandoned wells in western Pennsylvania of 12 kg d–1 (n = 7) and 550 g d–1 (n = 42), respectively, compared to 360 g d–1 (n = 35) and 530 g d–1 (n = 53), respectively, measured by Kang et al.27,28 (Table S6). The mean value for plugged wells in this investigation was dominated by one well (059-01162) having a methane emission rate of 83 kg d–1 (Figure S14). In the absence of this well, the mean emission rate for plugged wells would be 390 g d–1 similar to that measured by Kang et al.27,28 At OCSP and HSP, we observed mean emission rates for unplugged wells at 54 g d–1 (n = 7) and 285 g d–1 (n = 11) compared to 27 g d–1 and (n = 129) and 700 g d–1 (n = 22) observed by Saint-Vincent et al.33 and Pekney et al.,30 respectively. Hence, our methane emission rate measurements are in general agreement with others conducting similar measurements in western Pennsylvania. We observed median methane emission rates for plugged and unplugged abandoned wells of 36 and 74 g d–1, respectively. As observed by others,29,32,34−36 methane emissions from both plugged and unplugged wells follow a “long tail distribution” where measurements from a few wells dominate total emissions as reflected in mean values being much higher than median values.

Mean methane emission rates from abandoned wells in Pennsylvania appear to be lower than those measured through SCVF for active wells. Ingraffea et al.7 estimated a daily flow of 110,400 m3 d–1 of methane to the atmosphere from SCVF at 62,483 conventional, unconventional, and coal-bed active wells, equivalent to 1.16 kg d–1 per well at 298 K and 1 atmosphere. This is greater than a factor of two for mean methane emissions at unplugged abandoned wells measured in this investigation and by Kang et al.27,28

Emission rates of VOCs from abandoned wells in this investigation were dominated by hexane, heptane, and cyclohexane, followed by BTEX compounds (Table 2 and Figure S15). We observed KM mean emission rates of hexane, heptane, cyclohexane, and benzene at 0.72, 0.38, 0.16, and 0.022 g d–1, respectively. For active marginal or stripper wells (less than 1 barrel oil equivalent per day) in Ohio, Deighton et al.23 observed mean emission rates for hexane, heptane, cyclohexane, and benzene at 21.6, 16.9, 5.9, and 1.2 g d–1, respectively. Hence, the mean emission rate of benzene in marginal oil wells was ∼55× higher than that observed in this investigation. Median values of emission rates of hexane, heptane, cyclohexane, and benzene in our investigation were less than mean values due to a small proportion of abandoned wells having high emission rates. Summary statistics for emission rates of other compounds were heavily left-censored (64.6–95.8%) and hence should be viewed with caution.

In this investigation, benzene emission rates could not be measured at 24 of 48 (50%) abandoned wells because gas flow was too low to be measured (10 wells), benzene was not detected (3 wells), or both conditions were present (11 wells) (Figure 3). Also, emission of methane did not necessarily correspond to the emission of benzene. For instance, the plugged well with the highest flow rate and methane emission rate (059-01162) (83 kg d–1) had a non-measurable benzene emission rate. Hence, the emission of benzene cannot be assumed based on the presence of methane or a leakage flow rate from an abandoned well. Depending on the limit of flow and benzene detection at individual wells, the limits of measurement of benzene emission rates in this investigation varied from 3.0 × 10–6 to 8.5 × 10–3 g d–1.

Figure 3.

Figure 3

Open in a new tab

Emission rates for benzene when detectable flow and benzene concentrations are present, detectable concentrations of benzene, and detectable flow rates (plotted in this order) for unplugged and plugged abandoned wells in western Pennsylvania. Wells with both non-detectable flow and benzene concentrations are left blank.

3.4. Proximity Analysis

Of 18,608 documented abandoned wells with coordinates in Pennsylvania, a proximity analysis indicated that there are 499 (2.68%), 4243 (22.8%), and 17,299 (93.0%) abandoned wells within 10 m, 100 m, and 1 km of a building, respectively (Figure 4a). Our accuracy check of 5% of surveyed abandoned wells (n = 213) found a total of 92.5% (n = 197) true positives, 7.0% (n = 15) false positives, and 0.5% (n = 1) demolished (Table S7). While 5% of the wells surveyed are a relatively small subset of our dataset, given the random selection of these wells and their distribution across the entirety of our study area, we expect a similar distribution of true and false positives within the entire dataset.

Figure 4.

Figure 4

Open in a new tab

Histograms (bin size = 10 m) illustrating the proximity of documented abandoned wells (a) to the nearest building in Pennsylvania and (b) to the nearest residence in Allegheny County in western Pennsylvania. Wine and orange colored dashed lines represent setback distances from buildings in Pennsylvania for conventional (200 ft) and unconventional (500 ft) oil and gas wells (58 PA CS 3215).

Of 430 documented abandoned wells with coordinates in densely populated Allegheny County, there are 42 (9.77%), 176 (40.9%), and 373 (86.7%) abandoned wells within 10 m, 100 m, and 1 km of a residence, respectively (Figure 4b). Since the actual number of abandoned wells likely exceeds the number of documented abandoned wells by almost an order of magnitude and the percentage of false positives in our proximity analysis is relatively low, the actual number of abandoned wells within 10 or 100 m of buildings in Pennsylvania and residences in Allegheny County is likely much larger than presented here.

Further evidence of concern regarding the proximity of abandoned wells to buildings and residences is provided by field observations. During this investigation, the PADEP initiated an emergency plug and abandon action at 003-00832 because of complaints of odor by residents in a new housing development over 1 km from the well. Gas from this unplugged abandoned well had a benzene concentration of 250 ppmv and a benzene emission rate of 0.74 g d–1—the highest emission rate measured in this investigation. Acrid gas emissions at this well-made sampling were challenging, indicating that future field work should include respiratory protection. At one abandoned unplugged well (125-21260), we could not calculate a benzene emission rate due to high wind and difficulty enclosing leak locations. However, gas from this abandoned well had a benzene concentration of 212 ppmv and was located at the edge of a driveway adjacent to a house. At one location, an unplugged abandoned well was located in a backyard close (10–15 m) to a rented residence, where the renter was unaware that the pipe exiting the ground was an abandoned well. Although benzene was not detected in the gas from this well, 100% gas (i.e., 100% instrument response) was observed during screening, and methane was being emitted at a rate of 3.0 kg d–1 representing a potential fire hazard.

4. Conclusions

Gas emissions from abandoned oil and gas wells have recently received considerable interest due to their potential impacts on the climate. However, these studies have been limited to methane. This is the first study in which VOC concentrations and associated emission rates have been measured in abandoned wells. In this investigation, we demonstrate that: (1) gas from abandoned wells contains VOCs, including benzene, a known human carcinogen; (2) VOCs are emitted from abandoned wells, the magnitude of which depends on the flow rate and concentration of VOCs in the gas stream; and (3) abandoned wells are located in proximity to buildings and residences in western Pennsylvania. Together, these observations indicate that further investigation is necessary to determine whether gas emissions pose an inhalation risk to people living, working, or congregating near abandoned wells. Also, benzene concentrations measured in active and abandoned wells in this investigation in western Pennsylvania were lower than those found in active wells in other parts of the United States. Hence, benzene emissions from abandoned wells at other locations could be higher than those observed in this investigation, necessitating further investigation in other geographic areas.

Finally, in this investigation, we had a dual objective of determining both the concentration and emission rate of VOCs in gas from abandoned wells. Moving forward, though, given the difficulty in sample collection (air intrusion) and low flow measurement, if the primary objective is the determination of VOC emission rates, a flux-based methodology employed by Kang et al.26−28 and others29−36 should be utilized and supplemented with VOC analysis and equipment blanks.

Acknowledgments

The authors would like to acknowledge Rick Swank and James Braunns of the PADEP for locating and screening gas leakage at abandoned wells. The authors would also like to acknowledge Seth Pelepko and William Walsh of the PADEP for their assistance in planning sampling activities and gathering information on abandoned wells. The PADEP provided PFA and FEP gas sampling bags used during sampling. The authors would also like to acknowledge comments provided by three anonymous reviewers to improve the manuscript.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.3c00676.

  • Estimation of flowrates, sample locations, carbon-13 analysis, estimation of air intrusion into samples, proximity analysis, and VOCs in equipment blanks (PDF)

Author Contributions

Research concept origination: S.B.C.S. and D.D. Field screening, sample collection, data interpretation, and manuscript preparation: D.D. and R.R. Proximity analysis: R.R. Carbon-13 analysis and manuscript review: E.L. Manuscript review: K.B., D.M., and S.B.C.S.

This research was supported by the Energy Foundation (grant #56336) and The Heinz Endowments (grant #E8891).

The authors declare no competing financial interest.

Supplementary Material

ao3c00676_si_001.pdf (2.9MB, pdf)

References

  1. United States Environmental Protection Agency . Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2018 (Reports and Assessments EPA 430-R-20-002), 2020.
  2. PADEP (Pennsylvania Department of Environmental Protection) . DEP Office of Oil and Gas Management Orphan and Abandoned Wells. http://cedatareporting.pa.gov/Reportserver/Pages/ReportViewer.aspx?/Public/DEP/OG/SSRS/Abandoned_Orphan_Web (accessed Sept, 2022).
  3. Pelepko S.; Byron L. A.; Beattie s.; Catalano D.; Swank R.. Pennsylvania legacy well integrity and emissions study: Part 1. Joint 52nd Northeastern Annual Section/51st North-Central Annual Section Meeting—2017; Geological Society of America, 2017.
  4. Pennsylvania Department of Environmental Protection . Abandoned and Orphan Oil and Gas Wells and the Well Plugging Program. http://www.depgreenport.state.pa.us/elibrary/PDFProvider.ashx?action=PDFStream&docID=1419023&chksum=&revision=0&docName=ABANDONED+AND+ORPHAN+OIL+AND+GAS+WELLS+AND+THE+WELL+PLUGGING+PROGRAM&nativeExt=pdf&PromptToSave=False&Size=411528&ViewerMode=2&overlay=0 (accessed Jan 16, 2022).
  5. Pennsylvania Department of Conservation and Natural Resources. State Park Boundaries. https://www.pasda.psu.edu/uci/DataSummary.aspx?dataset=114 (accessed July 18, 2022).
  6. Pennsylvania Department of Conservation and Natural Resources. State Forest Lands. https://www.pasda.psu.edu/uci/DataSummary.aspx?dataset=263 (accessed July 18, 2022).
  7. Ingraffea A. R.; Wawrzynek P. A.; Santoro R.; Wells M. Reported methane emissions from active oil and gas wells in Pennsylvania, 2014-2018. Environ. Sci. Technol. 2020, 54, 5783–5789. 10.1021/acs.est.0c00863. [DOI] [PubMed] [Google Scholar]
  8. Lackey G.; Rajaram H. Modeling gas migration, sustained casing pressure, and surface casing vent flow in onshore oil and gas wells. Water Resour. Res. 2019, 55, 298–323. 10.1029/2018wr024066. [DOI] [Google Scholar]
  9. Abboud J. M.; Watson T. L.; Ryan M. C. Fugitive methane migration around Alberta’s petroleum wells. Greenhouse Gases: Sci. Technol. 2021, 11, 37–51. 10.1002/ghg.2029. [DOI] [Google Scholar]
  10. Wisen J.; Chesnaux R.; Werring J.; Wendling G.; Baudron P.; Barbecot F. A portrait of wellbore leakage in northeastern British Columbia, Canada. Proc. Natl. Acad. Sci. U.S.A. 2020, 117, 913–922. 10.1073/pnas.1817929116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Nowamooz A.; Lemieux J.-M.; Molson J.; Therrien R. Numerical investigation of methane and formation fluid leakage along the casing of a decommissioned shale gas well. Water Resour. Res. 2015, 51, 4592–4622. 10.1002/2014wr016146. [DOI] [Google Scholar]
  12. Pennsylvania Statue 58 § 4. https://casetext.com/statute/pennsylvania-statutes/statutes-unconsolidated/title-58-ps-oil-and-gas/chapter-1-oil-and-gas-wells/care-of-and-protection-of-wells-and-manner-of-operation-in-general/section-4-manner-of-plugging-abandoned-oil-and-gas-wells-vent-pipes-in-abandoned-wells-bored-through-coal (accessed July 6, 2022).
  13. Bachu S. Analysis of gas leakage occurrence along wells in Alberta, Canada, from a GHG perspective – Gas migration outside casing. Int. J. Greenh. Gas Control 2017, 61, 146–154. 10.1016/j.ijggc.2017.04.003. [DOI] [Google Scholar]
  14. Darrah T. H.; Vengosh A.; Jackson R. B.; Warner N. R.; Poreda R. J. Noble gases identify the mechanisms of fugitive gas contamination in drinking-water wells overlying the Marcellus and Barnett Shales. Proc. Natl. Acad. Sci. U.S.A. 2014, 111, 14076–14081. 10.1073/pnas.1322107111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Heilweil V. M.; Stolp B. J.; Kimball B. A.; Susong D. D.; Marston T. M.; Gardner P. M. A stream-based methane monitoring approach for evaluating groundwater impacts associated with unconventional gas development. Groundwater 2013, 51, 511–524. 10.1111/gwat.12079. [DOI] [PubMed] [Google Scholar]
  16. Heilweil V. M.; Grieve P. L.; Hynek S. A.; Brantley S. L.; Solomon D. K.; Risser D. W. Stream measurements locate thermogenic methane fluxes in groundwater discharge in an area of shale-gas development. Environ. Sci. Technol. 2015, 49, 4057–4065. 10.1021/es503882b. [DOI] [PubMed] [Google Scholar]
  17. Pelepko S.PIOGATECH Well Plugging Workshop: Well plugging regulatory viewpoint. Pennsylvania Independent Oil and Gas Association. April 25, 2019. https://pioga.org/wp-content/uploads/2019/05/Well-Plugging-Workshop_Pelepko-DEP.pdf. (accessed July 6, 2022).
  18. Pelepko S.CCAC Abandoned and Orphan Oil and Gas Wells in Pennsylvania, October 27, 2020. https://files.dep.state.pa.us/Energy/Office%20of%20Energy%20and%20Technology/OETDPortalFiles/Climate%20Change%20Advisory%20Committee/2020/10-27-20/Presentation_Abandoned%20Wells_Pelepko_20201027.pdf (accessed July 6, 2022).
  19. Pennsylvania Department of Environmental Protection . Stray Gas Migration Associated with Oil and Gas Wells (PA DEP Technical Advisory Board) DRAFT Report, 2009. http://files.dep.state.pa.us/OilGas/BOGM/BOGMPortalFiles/TechnicalAdvisoryBoard/2009/Stray%20Gas%20Migration%20Cases.pdf. (accessed July 6, 2022).
  20. Chilingar G. V.; Endres B. Environmental hazards posed by the Los Angeles Basin urban oilfields: an historical perspective of lessons learned. Environ. Geol. 2005, 47, 302–317. 10.1007/s00254-004-1159-0. [DOI] [Google Scholar]
  21. Dusseault M. B.; Gray M. N.; Nawrocki P. A.. Why Oilwells Leak: Cement behavior and Long-Term Consequences; Society of Petroleum Engineers, 2000, p 64733. [Google Scholar]
  22. Dusseault M.; Jackson R. Seepage pathway assessment for natural gas to shallow groundwater during well stimulation, in production, and after abandonment. Environ. Geosci. 2014, 21, 107–126. 10.1306/eg.04231414004. [DOI] [Google Scholar]
  23. Deighton J. A.; Townsend-Small A.; Sturmer S. J.; Hoschouer J.; Heldman L. Measurements show that marginal wells are a disproportionate source of methane relative to production. J. Air Waste Manage. Assoc. 2020, 70, 1030–1042. 10.1080/10962247.2020.1808115. [DOI] [PubMed] [Google Scholar]
  24. Lyman S. N.; Watkins C.; Jones C. P.; Mansfield M. L.; McKinley M.; Kenney D.; Evans J. Hydrocarbon and carbon dioxide fluxes from natural gas well pad soils and surrounding soils in eastern Utah. Environ. Sci. Technol. 2017, 51, 11625–11633. 10.1021/acs.est.7b03408. [DOI] [PubMed] [Google Scholar]
  25. El Hachem K. E.; Kang M. Methane and Hydrogen Sulfide Emissions from Abandoned, Active, and Marginally Producing Oil and Gas Wells in Ontario, Canada. Sci. Total Environ. 2022, 823, 153491. 10.1016/j.scitotenv.2022.153491. [DOI] [PubMed] [Google Scholar]
  26. Kang M.; Kanno C. M.; Reid M. C.; Zhang X.; Mauzerall D. L.; Celia M. A.; Chen Y.; Onstott T. C. Direct measurements of methane emissions from abandoned oil and gas wells in Pennsylvania. Proc. Natl. Acad. Sci. U.S.A. 2014, 111, 18173–18177. 10.1073/pnas.1408315111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Kang M.; Christian S.; Celia M. A.; Mauzerall D. L.; Bill M.; Miller A. R.; Chen Y.; Conrad M. E.; Darrah T. H.; Jackson R. B. Identification and characterization of high methane-emitting abandoned oil and gas wells. Proc. Natl. Acad. Sci. U.S.A. 2016, 113, 13636–13641. 10.1073/pnas.1605913113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kang M.; Christian S.; Celia M. A.; Mauzerall D. L.; Bille M.; Miller A. R.; Chen Y.; Conrad M. E.; Darrah T. H.; Jackson R. B. Correction for “Identification and characterization of high methane-emitting abandoned oil and gas wells. Proc. Natl. Acad. Sci. U.S.A. 2016, 113, 13636. 10.1073/pnas.1605913113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Lebel E. D.; Lu H. S.; Vielstädte L.; Kang M.; Banner P.; Fischer M. L.; Jackson R. B. Methane emissions from abandoned oil and gas wells in California. Environ. Sci. Technol. 2020, 54, 14617–14626. 10.1021/acs.est.0c05279. [DOI] [PubMed] [Google Scholar]
  30. Pekney N. J.; Diehl J. R.; Ruehl D.; Sams J.; Veloski G.; Patel A.; Schmidt C.; Card T. Measurement of methane emissions from abandoned oil and gas wells in Hillman State Park, Pennsylvania. Carbon Manag. 2018, 9, 165–175. 10.1080/17583004.2018.1443642. [DOI] [Google Scholar]
  31. Riddick S. N.; Mauzerall D. L.; Celia M. A.; Kang M.; Bressler K.; Chu C. D.; Gum C. D. Measuring methane emissions from abandoned and active oil and gas wells in West Virginia. Sci. Total Environ. 2019, 651, 1849–1856. 10.1016/j.scitotenv.2018.10.082. [DOI] [PubMed] [Google Scholar]
  32. Saint-Vincent P. M. B.; Reeder M. D.; Sams J. I. III; Pekney N. J. An Analysis of abandoned oil well characteristics affecting methane emissions in the Cherokee Platform in Eastern Oklahoma. Geophys. Res. Lett. 2020, 47, e2020GL089663 10.1029/2020GL089663. [DOI] [Google Scholar]
  33. Saint-Vincent P. M. B.; Sams J. I. III; Reeder M. D.; Mundia-Howe M.; Veloski G. A.; Pekney N. J. Historic and modern approaches for discovery of abandoned wells for methane emissions mitigation in Oil Creek State Park, Pennsylvania. J. Environ. Manage. 2021, 280, 111856. 10.1016/j.jenvman.2020.111856. [DOI] [PubMed] [Google Scholar]
  34. Townsend-Small A.; Ferrara T. W.; Lyon D. R.; Fries A. E.; Lamb B. K. Emissions of coalbed and natural gas methane from abandoned oil and gas wells in the United States. Geophys. Res. Lett. 2016, 43, 2283–2290. 10.1002/2015gl067623. [DOI] [Google Scholar]
  35. Townsend-Small A.; Hoschouer J. Direct measurements from shut-in and other abandoned wells in the Permian Basin of Texas indicate some wells are a major source of methane emissions and produced water. Environ. Res. Lett. 2021, 16, 054081. 10.1088/1748-9326/abf06f. [DOI] [Google Scholar]
  36. Williams J. P.; Regehr A.; Kang M. Methane emissions from abandoned oil and gas wells in Canada and the United States. Environ. Sci. Technol. 2021, 55, 563–570. 10.1021/acs.est.0c04265. [DOI] [PubMed] [Google Scholar]
  37. United States Geological Survey Energy Resources Program Geochemistry Laboratory Database (EGDB). https://certmapper.cr.usgs.gov/data/apps/geochem-db/ (accessed Jan 31, 2022).
  38. Colorado Oil and Gas Conservation Commission. Oil and Gas Well Analytical Data Download. https://cogcc.state.co.us/documents/data/downloads/environmental/ProdWellDownLoad.html (accessed April 4, 2022).
  39. DiGiulio D. C.; Jackson R. B. Impact to Underground Sources of Drinking Water and domestic wells from production well stimulation and completion practices in the Pavillion, Wyoming Field. Environ. Sci. Technol. 2016, 50, 4524–4536. 10.1021/acs.est.5b04970. [DOI] [PubMed] [Google Scholar]
  40. Eastern Research Group, Inc. Condensate Tank Oil and Gas Activities, Final Report, Prepared for: Texas Commission on Environmental Quality, Air Quality Division, October 10, 2012. https://www.tceq.texas.gov/assets/public/implementation/air/am/contracts/reports/ei/5821199776FY1211-20121031-ergi-condensate_tank.pdf (accessed Jan 31, 2022).
  41. United States Environmental Protection Agency . Benzene. https://www.epa.gov/sites/default/files/2016-09/documents/benzene.pdf (accessed July 6, 2022).
  42. California Office of Environmental Health Hazard Assessment . OEHHA 2008. Technical Supporting Document for Noncancer RELs, Appendix D1, 2014. https://oehha.ca.gov/media/downloads/crnr/appendixd1final.pdf (accessed July 6, 2022).
  43. Bascom-Turner Unit Specifications https://www.bascomturner.com/_files/ugd/9a5810_2b30bb86858742ce8e9b64128837a4e3.pdf (accessed Aug 1, 2022).
  44. United States Environmental Protection Agency . Protocol for Equipment Leak Emissions Estimates. Office of Air Quality Planning and Standards, Research Triangle Park, NC 27711, EPA-453/R-95-017, November, 1995.
  45. Hayes H. C.; Benton D. J.; Khan N. Impact of Sampling Media on Soil Gas Measurements. A WMA Vap. Intrusion Next Great Environ. Challenges 2006, 13. [Google Scholar]
  46. Mussato B.; Varisco S.; Tsurikova L.. Assessment of Volatile Organic Compounds from Tubing Used for Soil Vapour Sample Collection. Prepared by: CARO Analytical Services. CARO Anal. Serv. p. Submitted to: Science Advisory Board for Contaminated Sites in British Columbia Department of Chemistry University of Victoria. April 1, 2009. Vol 17. [Google Scholar]
  47. Laughrey C. D.; Baldassare F. J. Geochemistry and origin of some natural gases in the Plateau Province, Central Appalachian Basin, Pennsylvania and Ohio. Am. Assoc. Pet. Geol. 1998, 82, 317–335. 10.1306/1D9BC403-172D-11D7-8645000102C1865D. [DOI] [Google Scholar]
  48. Laughrey C. D.; Billman D. A.; Canich M. R. Petroleum geology and geochemistry of the Council Run gas field, north central Pennsylvania. Am. Assoc. Pet. Geol. 2004, 88, 213–239. 10.1306/10060301104. [DOI] [Google Scholar]
  49. Breen K. J.; Révész K.; Baldassare F. J.; McAuley S. D.. Natural Gases in Ground Water Near Tioga Junction, Tioga County, North-Central Pennsylvania—Occurrence and Use of Isotopes to Determine Origins Report 2007-5085; U.S. Geological Survey Scientific Investigations, 2005.
  50. Wood M. D.; Beresford N. A.; Copplestone D. Limit of detection values in data analysis: Do they matter?. Radioprotection 2011, 46, S85–S90. 10.1051/radiopro/20116728s. [DOI] [Google Scholar]
  51. Helsel D. R.Nondetects and Data Analysis. Statistics for Censored Environmental Data; John Wiley and Sons: NY, 2005. [Google Scholar]
  52. Baccarelli A.; Pfeiffer R.; Consonni D.; Pesatori A. C.; Bonzini M.; Patterson D. G. Jr; Bertazzi P. A.; Landi M. T. Handling of dioxin measurement data in the presence of non-detectable values: Overview of available methods and their application in the Seveso chloracne study. Chemosphere 2005, 60, 898–906. 10.1016/j.chemosphere.2005.01.055. [DOI] [PubMed] [Google Scholar]
  53. U.S. Environmental Protection Agency . ProUCL Version 5.2.0 Technical Guide Statistical Software for Environmental Applications for Data Sets with and without Nondetect Observations, version 5.2.0 prepared by; Neptune and Company, Inc.: 1435 Garrison Street, Suite 201 Lakewood, CO 80215, 2022.
  54. Kaplan E. L.; Meier P. Nonparametric estimation from incomplete observations. J. Am. Stat. Assoc. 1958, 53, 457–481. 10.1080/01621459.1958.10501452. [DOI] [Google Scholar]
  55. Antweiler R. C.; Taylor H. E. Evaluation of statistical treatments of left-censored environmental data using coincident uncensored data sets: I. Summary statistics. Environ. Sci. Technol. 2008, 42, 3732–3738. 10.1021/es071301c. [DOI] [PubMed] [Google Scholar]
  56. Tarone R.; Ware J. On distribution-free tests for equality of survival sistributions. Biometrika 1977, 64, 156–160. 10.1093/biomet/64.1.156. [DOI] [Google Scholar]
  57. Microsoft Maps. US Building Footprints, 2021. https://github.com/Microsoft/USBuildingFootprints (accessed on Jan 15, 2023).
  58. Simple Features for R. [R]. r-spatial, 2022. https://github.com/r-spatial/sf (accessed on Jan 15, 2023) (Original work published 2015).
  59. Allegheny County. Allegheny County Property Assessments, 2021. https://data.wprdc.org/dataset/property-assessments (accessed on Jan 15, 2023).
  60. R Core Team . R: A language and environment for statistical computing (4.0.3); R Foundation for Statistical Computing, 2020. https://www.R-project.org/ (accessed on Jan 15, 2023).
  61. Allegheny County Division of Computer Services Geographic Information Systems Group . Allegheny County Parcels 202112, 2021. https://www.pasda.psu.edu/uci/DataSummary.aspx?dataset=1214 (accessed on Jan 15, 2023).
  62. Surdex Corporation . National Agricultural Imagery Program (0.6 meter resolution) 2019, 2019. https://www.pasda.psu.edu/uci/DataSummary.aspx?dataset=3211 (accessed on Jan 15, 2023).
  63. Quantum Spatial, Inc. . PEMA 2018 0.5-foot Orthoimagery—Cycle 1 2018, 2018. https://www.pasda.psu.edu/uci/DataSummary.aspx?dataset=5158 (accessed on Jan 15, 2023).
  64. Google Earth Pro (7.3.4.8248), 2021.
  65. Boothroyd I. M.; Almond S.; Qassim S. M.; Worrall F.; Davies R. J. Fugitive emissions of methane from abandoned, decommissioned oil and gas wells. Sci. Total Environ. 2016, 547, 461–469. 10.1016/j.scitotenv.2015.12.096. [DOI] [PubMed] [Google Scholar]
  66. British Columbia Oil and Gas Commission . Gas Migration Investigation Report, 2013. https://www.bcogc.ca/node/14620/(accessed July 6, 2022).
  67. Fleming N. A.; Morais T. A.; Mayer K. U.; Ryan M. C. Spatiotemporal variability of fugitive gas migration emissions around a petroleum well. Atmos. Pollut. Res. 2021, 12, 101094. 10.1016/j.apr.2021.101094. [DOI] [Google Scholar]
  68. Forde O. N.; Cahill A. G.; Beckie R. D.; Mayer K. U. Barometric-pumping controls fugitive gas emissions from a vadose zone natural gas release. Sci. Rep. 2019, 9, 14080. 10.1038/s41598-019-50426-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Samano P. S. G.; Cahill A. G.; Timmis R.; Busch A. Constraining well integrity and propensity for fugitive gas migration in surficial soils at onshore decommissioned oil and gas wells in England. Int. J. Greenh. Gas Control 2022, 119, 103712. 10.1016/j.ijggc.2022.103712. [DOI] [Google Scholar]
  70. Sechman H. Detailed analysis of gaseous components in soil gases around petroleum wells – An effective tool for evaluation of their integrity. Appl. Geochem. 2022, 142, 105346. 10.1016/j.apgeochem.2022.105346. [DOI] [Google Scholar]
  71. Yurkevich N.; Michurin E.; Yurkevich N. Continuous monitoring system for soil gas migration in leaking Oil and Gas wells. IOP Conf. Series. Earth Environ. Sci. 2019, 377, 012055. 10.1088/1755-1315/377/1/012055. [DOI] [Google Scholar]
  72. Cahill A. G.; Steelman C. M.; Forde O.; Kuloyo O.; Emil Ruff S.; Mayer B.; Mayer K. U.; Strous M.; Ryan M. C.; Cherry J. A.; Parker B. L. Mobility and persistence of methane in groundwater in a controlled-release field experiment. Nat. Geosci. 2017, 10, 289–294. 10.1038/ngeo2919. [DOI] [Google Scholar]
  73. Forde O. N.; Mayer K. U.; Cahill A. G.; Mayer B.; Cherry J. A.; Parker B. L. Vadose zone gas migration and surface effluxes after a controlled natural gas release into an unconfined shallow aquifer. Vadose Zone J. 2018, 17, 1–16. 10.2136/vzj2018.02.0033. [DOI] [Google Scholar]
  74. Yin Y.; Mastalerz M.; Lennon J. T.; Drobniak A.; Schimmelmann A. Characterization and microbial mitigation of fugitive methane emissions from oil and gas wells: Examle from Indiana, USA. Appl. Geochem. 2020, 118, 104619. 10.1016/j.apgeochem.2020.104619. [DOI] [Google Scholar]
  75. Jenden P. D.; Drazan D. J.; Kaplan I. R.. Mixing of Thermogenic Natural Gases in Northern Appalachian Basin; The American Association of Petroleum Geologist Bulletin, 1993, pp 980–998. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

ao3c00676_si_001.pdf (2.9MB, pdf)

Articles from ACS Omega are provided here courtesy of American Chemical Society

RESOURCES