Characterizing methane emissions from orphaned coalbed methane wells in the powder river basin

In 2018, the U.S. EPA Greenhouse Gas Inventory (GHGI) began including methane emissions from abandoned oil and gas (AOG) wells and estimated that they may constitute up to 4% of total methane emissions from the oil and gas sector. Several studies have shown that these emissions vary by location which introduces regionally dependent uncertainty into inventory estimates. In Wyoming, there are over 1000 wells identified as ‘orphaned’ indicating that they are both abandoned and unplugged, approximately 80% of which are coalbed methane (CBM) wells. In this pilot study, 3-hour measurements of ten orphaned CBM wells were taken to support the characterization of AOG well emission rates in the Powder River Basin (PRB) in Wyoming. The mean methane emission rate measured at these wells was 653 mg CH4/hr with the majority of these wells emitting in the 100-1000mg CH4/hr range. The highest emitting well in the study was measured at 4.53 g CH4/hr. Compared to the GHGI AOG wells emission factor of 10.04 g CH4/hr, these findings suggest that AOG wells emission rate in the PRB are overestimated by national inventory methods. Finally, the addition of AOG well emissions data from this study doubles the number of sampled AOG wells in Wyoming and triples the number of sampled CBM wells in the U.S.


Introduction
Several studies from the past decade have shown that Oil and Gas (O&G) wells may continue to emit methane after production has ceased ( (CH 4 ) is a powerful greenhouse gas (GHG) with a global warming potential 29.8 (±11) times that of carbon dioxide (CO 2 ) over a period of 100 years, and controlling fugitive CH 4 emissions is an effective near-term pathway to mitigate the effects of climate change (Forster et al 2021, Lecocq et al 2021. To that end, the U.S. Methane Emissions Reduction Action Plan (USMRP) lists plugging abandoned oil and gas (AOG) wells as one of five key points for reducing methane emissions in the O&G sector (Office of Domestic Climate Policy 2021). In line with the USMRP, the Bipartisan Infrastructure Law (BIL) that passed in late 2021 allocates $4.7B for orphaned well site plugging, remediation, and restoration activities (H.R.3684 -Infrastructure Investment and Jobs Act 2021).
There have been increased efforts in the past decade to quantify methane emissions from AOG wells for incorporation into GHG inventories. As recently as 2018, the U.S. Environmental Protection Agency (EPA) began including emissions estimates from AOG wells to its annual Inventory of U.S. Greenhouse Gas Emissions and Sinks (GHGI). The 2020 GHGI estimates that AOG wells emitted 276 kilotons of CH 4 , or 6.9 million metric tons of CO 2 equivalent (CO 2 e), which is approximately 3.3% of O&G methane emissions (EPA 2022). According to the EPA (2018), these estimates were derived using emission factors (EF) based on three studies: Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. Kang et al (2014Kang et al ( , 2016, Townsend-Small et al (2016). One set of EFs were applied to wells categorized geographically as 'Appalachian Basin' and another for 'entire U.S.' A recent survey of AOG well studies by Williams et al (2021) has shown that emissions can vary drastically by geographic location, emphasizing the need for AOG well quantification by basin. To date, only one study (Townsend-Small et al 2016) in the body of AOG well emissions literature measured AOG wells in Wyoming, which were all located on public land. Within their Wyoming sample of ten plugged wells, with five identified as coalbed methane (CBM), no detectable emissions were reported for any of the wells. This report focuses on 10 orphaned and mechanically shut-in CBM wells located on private land in the Powder River Basin (PRB) in Wyoming that were sampled during a pilot study in December, 2020. The Powder River Basin lies in Northeastern Wyoming and Southeastern Montana (shown in figure 1), with the majority of the basin in Wyoming, and is the largest coal deposit in the U.S. (Luppens et al 2015). The data from this study adds to the growing body of AOG wells literature and may serve as a reference for abandoned CBM well emissions in the PRB.

Coalbed methane
Unlike conventional gas reservoirs, methane in coal is not stored as free gas, but rather as adsorbed gas, at nearliquid densities on the internal surface area of the microporous coal (Puri and Yee 1990). As absolute-or partialpressure of methane is lowered during CBM production operations, gas molecules are desorbed from the coals internal surfaces and travel by diffusion through the coal matrix to the cleat system where they are conveyed according to Darcys law to a producing well (Gamson et al 1996). This is typically done by removing water with an electronic submersible pump from the well bore to lower the total pressure (Jenkins et al 2006). Figure 2 is a diagram showing the movement of gas molecules through the matrix to the fracture or cleat system.
The capacity of a coal matrix to store gas is a function of pressure and described by the Langmuir adsorption isotherm equation, where V is the volume of the adsorbed gas in the coal matrix, p is the pressure of the system, V L is the Langmuir volume constant representing the maximum volume of gas that can be adsorbed, and p L is the Langmuir pressure constant representing the pressure at which the storage capacity of the coal is equal to one-half the Langmuir volume (V L ) (Langmuir 1916, Mavor et al 1990. Figure 3 shows an average methane adsorption isotherm for a Powder River Basin coal. This figure shows that when pressure is high, the rate of gas desorption is relatively low (flatter portion of the curve), but as pressure becomes lower, the rate of gas desorption (and production) increases. Initially, when a coal seam is breached by drilling, there may be little, if any, gas production because the coal might not be fully saturated with methane. As pressure continues to be lowered by the production of water from the coal cleats, gas will begin to be desorbed and produced at larger rates. Ultimately, gas production will tail off as the differential pressure between the internal coal matrix and the producing well becomes too small to economically produce the gas. At that point, the well is shut in, plugged with cement, and abandoned.

Orphan well plugging in Wyoming
When O&G companies retire a well, abatement procedures according to most O&G producing states' statutes include the removal of surface equipment and plugging the wellbore. The Wyoming Oil and Gas Conservation  Commission (WOGCC) defines plugging rules specific to CBM wells such as plugging cement density, plug material makeup, and plugging length intervals along the well casing (055-3 Wyo. Code R. section 3-18, (WOGCC, n.d.-a)). The WOGCC defines orphaned wells as 'wells for which the agency is unable to require the responsible party (Owner or Operator) to plug and abandon them and rehabilitate the surface because the responsible party no longer operates in the state, is bankrupt, or is out-of-business' (WOGCC, n.d.-b). Once a well is orphaned in Wyoming, the plugging and reclamation responsibilities fall on the WOGCC. Many of the orphaned wells in the PRB are a result of the 2000s boom and bust of CBM (Bills Walsh 2017). U.S. CBM production peaked in 2008 when it made up 8% of total U.S. natural gas production, a third of which came from the PRB (EIA, 2022a, 2022b). During the sampling period of this pilot study, there were approximately 1900 orphaned wells in Wyoming. In late 2019, 88% of Wyoming's orphaned wells were CBM, and by June 2022, that percentage dropped to 71% (WOGCC, n.d.-c). The subsurface depth of the Big George coal seam is shallow enough to allow CBM well drilling to be done by modified water-well rigs (Jenkins et al 2006). As a result of their shallow depths, PRB CBM wells are more economical to plug than conventional wells, which explains the disproportional CBM plugging rate in Wyoming. The cost advantage of plugging CBM wells is evident by the estimated plugging costs in Wyoming ranging $5,000-$7,000 (WOGCC, n.d.-b), whereas the median cost of plugging is elsewhere in the U.S. ranges from $11,000 -$20,000, and up to $75,000 when including surface remediation (Kang et al 2019, Raimi et al 2021).

Well selection
For this pilot study, a University of Wyoming School of Energy Resources energy partner provided one week of escorted access to a number of abandoned wells on private land in the PRB located between Douglas, WY and Gillette, WY. Each of the candidate wells were completed in the Big George coal seam, one of the thickest coal seams in the PRB, with target depth between 750 ft and 1000 ft subsurface (Flores 2004). Eight of the wells in the study sample were open hole completions and the two remaining wells in the sample were cased and perforated completions. The list of accessible wells were exclusively those identified as orphaned in the WOGCC database (WOGCC, n.d.-c). Because these wells were on private land, many had existing box enclosures (shown in figure 4) for the safety of livestock. An experimental setup incorporating these enclosures, described later in section 3.3, was discussed with the energy partner prior to well selection. Therefore, only candidate wells with existing box enclosures were chosen by the energy partner. Finally, while in the field, wells were selected by the field manager based on ease of access for the University of Wyoming Mobile Research Laboratory considering the extreme winter weather conditions in the PRB.

Instruments
Two Picarro cavity ring-down spectrometers (CRDS) were employed to measure the escaping gas from the well and the background methane concentration simultaneously (models G2204 and G2401-m). Each model records at slightly different frequencies with the G2204 reporting at 2 Hz and the G2401-m reporting at 0.2 Hz. These CRDS instruments measure methane concentrations with a precision of±2 ppb.

Dynamic chamber sampling
Emissions estimates were obtained using a dynamic chamber approach (Townsend-Small et al 2016, Pekney et al 2018, Riddick et al 2019. Ambient air was drawn into the existing wellhead box enclosures at a controlled flow rate and the exiting air was continuously sampled by a CRDS to obtain a methane concentration. The volumetric flux, F, was calculated using the equation, where v is the flow rate (m 3 s −1 ), C Eq is the volumetric mixing ratio of CH 4 at steady state (ppmv), and C bkg is the average background methane volumetric mixing ratio (ppmv). The calculated fluxes were then converted to mass emission rates (mg CH 4 /hr) using the recorded temperature and pressure measurements at the flow meter. The average background concentration was subtracted from the instantaneous enclosure concentration data to observe any temporal variation in the emissions. Typical time to one full air change of the 4 ft×4 ft×4 ft box enclosures was calculated to be approximately 30 mins for a 60 LPM at STP (STP: P = 1 atm, T = 0°C). Hence, the first 30 mins of sampling were not used in the flux calculation to allow the enclosure to fully flush existing methane concentrations. Riddick et al (2020) noted that temporal variability in emissions introduced large uncertainty when measuring AOG wells and that high-emissions events tended to be short-lived (<1 h). Therefore, reported emission rates in this study were obtained by averaging over a continuous measurement of up to 3 h at each well as a balance between length of measurement and total time for measurement. Ultimately, this sampling length limited the feasible number of sites measured in a day.
A combination of plastic tarp and sandbags were used to seal the edges of the enclosure to ensure that flow into the enclosure was primarily from the drilled inlet holes. Additionally, two portable, intrinsically safe fans were placed inside the wellhead enclosure to ensure adequate mixing of inlet air and wellhead emissions. Additional details of the experimental setup are contained in the Supplemental Information of this document.

Emission factors
The EPA applies a status-dependent emission factor (i.e., plugged, unplugged) to regional well counts for estimating AOG well emissions in the GHGI (EPA 2018). These emission factors are derived using the mean emissions estimate from regional AOG well studies. For the studies with readily available unplugged AOG well emissions data (Townsend-Small et al 2016, Lebel et al 2020, Saint-Vincent et al 2020, Townsend-Small and Hoschouer 2021, Williams et al 2021 including this one, mean well emissions were bootstrap sampled to obtain a point estimate emission factor and 95% confidence interval (CI) for study intercomparison. The reported mean and 95% upper confidence limit from Riddick et al (2019) were utilized directly, and the 95% CI for Kang et al (2016) was inferred using the reported mean and standard error. Additional details of the bootstrap sampling can be found in the Supplemental Information.

Results
Individual emissions estimates for the ten sampled CBM wells are shown in table 1. The mean measured emission rate is 653 mg CH 4 /hr with most emission rates occurring in the 100-1000mg CH 4 /hr range. No significant linear relationships were found when exploring Pearson correlation coefficients between well emissions and well characteristics such as age (ρ = 0.26), well depth (ρ = − 0.38), hole diameter (ρ = 0.25), and cumulative gas production (ρ = 0.08).
The emission rates in table 1 were bootstrap sampled to obtain a 95% CI of [116.9, 1566.3] mg CH 4 /hr around a mean of 653 mg CH 4 /hr. Figure 5 shows this study's mean emission estimate and CI in comparison with previous studies. The previous studies' mean emission rates, including the GHGI AOG unplugged well 'Entire U.S.' EF, and standard errors were compared to this study's mean emission rate using the Welch's t-test for unequal variances resulting in p-values smaller than α = 0.05. Hence, this study's mean emission estimate significantly differs from previous studies.
The average background concentration was subtracted from the instantaneous enclosure concentration data to observe any temporal variation in the emissions. The background concentrations measured throughout the sampling days, typically between 6 AM to 6 PM, ranged from 2.01 to 2.29 ppm CH 4 , and the standard deviation of the background concentrations during individual site sampling periods ranged from <1 ppb to 100 ppb indicating that the background concentration was generally quite stable during these measurements and not the main driver of any observed variability in the calculated emission. Considering a background noise of 100 ppb with a sampling flow of 60 LPM flow rate at STP, the estimated minimum detection limit (MDL) during the field campaign was 0.258 mg CH 4 /hr.
Because this was a pilot study, we designed the experiment around the existing box enclosures while keeping the amount of extra equipment to a minimum. Therefore, we did not have extra equipment available in the field to measure soil fluxes. However, Jakel (1996) took measurements of methane ground seepage in the PRB and found an average soil flux of 0.03 mg m −2 day −1 . For our box enclosure footprint of 16 sq. ft., this is about 0.002 mg CH 4 /hr, which is well below the MDL calculated above. Furthermore, we are confident that all of the quantified emission rates were not significantly impacted by the calculated MDL and expected soils fluxes in this region.

Discussion
This study's mean emission estimate differs from previous studies' estimates, including the EPA GHGI AOG unplugged wells 'Entire U.S.' emission factor, highlighting the need for more localized EFs when including AOG  Combining their data with the data presented in this study, there is now a total of 10 unplugged AOG wells and 10 plugged AOG wells sampled in the state of Wyoming. One possible reason for the difference in EFs in the PRB versus previous studies elsewhere is the age of the PRB wells. According to WOGCC records, the oldest well sampled in this study was drilled in 2003. The age of these wells could mean that the well casings and above ground equipment remain structurally sound and, similar to findings by Townsend-Small and Hoschouer (2021), being shut-in is effectively preventing fugitive emissions. Although, we should point out that the data from this study and previous studies suggests that well age is a poor predictor of emissions (Saint-Vincent et al 2020, El Hachem andKang 2022).
Another reason for the difference may be the method of extraction. As noted in section 2.1, a CBM well is initially water-saturated which keeps the methane adsorbed to the coal surface. After a CBM is retired and water has stopped being produced to the surface, meteoric water can naturally re-saturate the coal bed raising the total pressure and ceasing the desorption of methane. This scenario is plausible since the Big George coal seam is an active aquifer and is fully saturated within one mile from the outcrop (Cook 2002). As an active aquifer, water in the Big George coal seam flows by gravity from the eastern outcrop downdip to the west to fill any void created by the pumping of water from the seam during CBM operations. Infiltration rates are slower than pumping out rates, but once pumping of water is stopped, water will begin to re-saturate the coal seam. Over time, the seam will be completely re-saturated with water and attempts to produce these wells again for methane may result in the need to produce and dispose of an additional volume of water.

Conclusion
The authors recognize that 10 samples is not sufficient to accurately characterize all AOG wells in the state of Wyoming, nor unplugged CBM well emissions for the entire U.S. However, the addition of AOG well emissions data from this study doubles the number of sampled AOG wells in Wyoming and triples the number of sampled CBM wells in the U.S. Although the number of orphaned wells in Wyoming is small in comparison to the estimated 130,000 orphaned wells in the U.S. (EPA 2022), the active industry of orphan well plugging in Wyoming is expanding and creating approximately 300 O&G jobs thanks to funding from the BIL (Bleizeffer 2022). These additional jobs are not insignificant considering that Wyoming's population is the lowest in the U.S. and that more than 70% of Wyoming state revenue comes from the fossil fuel industry (WY Department of Revenue 2022). We believe that the results of this study are especially useful for informing CBM well reclamation efforts specific to the PRB and Wyoming.
We showed that this study's sampled mean emission significantly differed from the EPA GHGI emission factor for unplugged wells in this region. This disagreement is a source of uncertainty in the national inventory, and future efforts should focus on the development of basin-specific EFs. The typical time required for equipment setup, methane sampling, and equipment tear-down was approximately 4 hours for each well sampled in this study. Although this technique is very precise, a faster measurement technique, such as the one used in Townsend-Small and Hoschouer (2021), is recommended for quickly improving the spatial characterization of AOG well emissions across the U.S. Finally, as the physical processes that govern the release of methane in CBM extraction are distinct from conventional O&G wells, we have presented a theory in section 5 that may explain the low methane emission rate from PRB CBM wells. Therefore, additional hydrogeological consideration should be given to EFs applied to abandoned CBM wells in future inventories.
In addition, our findings indicate that there may be important regional differences that impact emissions from AOG wells. This is particularly important when considering that well plugging is often handled at the state level. We have framed our work to motivate more regional investigations of AOG emissions, including by well type, that can provide regionally specific information to help limited state resources be allocated where it can have the biggest impact (e.g. to the highest emitting AOG wells).