Background

Geologic Energy Storage

The storage of energy for future use accommodates seasonal demand, allows for strategic stockpiles, and provides baseload power when renewable energy sources are variable. Energy consumption in the United States is shifting away from traditional fossil fuels like coal and oil and toward the inclusion of natural gas and renewable sources (U.S. Energy Information Administration [EIA], 2023a). Geologic energy storage refers to the storage of excess energy in underground settings. Above-ground energy storage often occurs in batteries and tanks, whereas underground storage in geologic formations allows for the retention of much greater quantities of energy—especially compressed gases and fluids—over much longer periods (Crotogino and others, 2018; Matos and others, 2019). Increased energy storage can benefit from an approach that encompasses all available storage types, including underground and above-ground technologies (Comello and Reichelstein, 2019). Underground storage affords flexibility, as different types and configurations of geologic formations can host different forms of energy.

In 2023, the U.S. Geological Survey (USGS) released Fact Sheet 2022–3082, “Geologic Energy Storage” (Buursink and others, 2023), which describes geologic energy storage and potential methods and settings for underground storage. Three broad categories of geologic storage, which include chemical, mechanical, and thermal methods, are described. The possible storage settings discussed are depleted hydrocarbon reservoirs, abandoned mines, purpose-drilled shafts, solution-mined salt caverns, and nonpotable aquifers (fig. 1) (Buursink and others, 2023). The work was motivated by the National Academies of Sciences, Engineering, and Medicine (NASEM) report “Future Directions for the U.S. Geological Survey's Energy Resources Program” (NASEM, 2018).

The NASEM report found that “Assessing the storage potential for various basins in the United States could become a new and strategically important priority for the ERP [USGS Energy Resources Program]” (NASEM, 2018, p. 101). The assessment methodology detailed in this report focuses on estimating capacities for storing natural gas and potentially storing other gaseous energy sources, such as hydrogen (using chemical storage methods), in depleted reservoirs that primarily produced natural gas. Other approaches being designed at the USGS focus on different combinations of storage methods and underground settings.

Assessment Methodology Scope

The methodology presented in this report examines natural gas storage in depleted hydrocarbon reservoirs in U.S. sedimentary basins. Methods for storing other energy gases, such as hydrogen, may be derived using additional steps that involve fluid substitution (refer to Lackey and others [2023] and Jones and others [2024] for further information). For this assessment, a gas reservoir is defined as any hydrocarbon accumulation with a cumulative produced gas-to-oil ratio (GOR) above 10,000 standard cubic feet of gas per barrel of oil (SCF/Bbl) (which is a conservative definition according to Carolus and others [2018] and Warwick and others [2019]). While the GOR of a reservoir can change during its production lifecycle, the cumulative production ratio is considered a fair representation for this assessment.

This report does not describe assessing gas storage in depleted oil reservoirs, aquifers, or salt caverns, because these features comprise a smaller resource base (Pipeline and Hazardous Materials Safety Administration, 2023). The storage of natural gas in depleted oil reservoirs also involves uncertainty, given the complicated gas and fluid phase interactions and the potential for contaminating stored gas, which is commonly encountered by storage operators (Plaat, 2009). Buoyant aquifer storage is not addressed due to the technical and geologic risks, the expense of characterizing the potential interaction with groundwater, and a lack of new U.S. projects of this type in recent decades (Schultz and Evans, 2020). However, storing gas, compressed air, and other fluids in salt caverns is relevant and could be addressed in future assessments.

Objective

The objective of this report is to describe an assessment methodology for estimating the volumes of natural gas, and possibly other energy gases, that could be stored in depleted hydrocarbon reservoirs using current technologies. The development of this methodology is guided by (1) established engineering practices used by gas storage facility operators to maximize seasonal storage capacity, (2) State regulations, and (3) economic factors (Coats, 1966; Katz and Coats, 1968).

The approach in this report does not account for all of the real-world conditions and external constraints under which natural gas storage facilities operate. Reservoir-specific engineering considerations not addressed by this assessment include leakage from abandoned wells, groundwater infiltration, and the compaction of pore space after reservoir production (for example, Chang and others [2014] and Michanowicz and others [2017]). The estimation process described here does not include economic or technical constraints, such as the availability of pipelines for gas transmission and limits on surface ownership related to accessing injection and withdrawal wells. In the “Demonstrating Storage Capacity in the Michigan Basin” section of this report, example assessment results are aggregated and reported for two plays in the Michigan Basin and show a first order estimate of additional gas storage capacity to demonstrate the application of this probabilistic methodology.

U.S. Geological Survey Assessment Background

The USGS periodically assesses major stratigraphic intervals in petroliferous basins to evaluate the potential volumes of undiscovered, technically recoverable, conventional, and continuous oil and gas resources onshore in the United States and in State waters (for examples, refer to Gautier and others [1996], Charpentier and Klett [2005], and the U.S. Geological Survey [2024]). The USGS also released methods and assessments of buoyant and residual trapping for the geologic sequestration of carbon dioxide (CO₂) (Brennan and others, 2010; Blondes and others, 2013a; U.S. Geological Survey Geologic Carbon Dioxide Storage Resources Assessment Team, 2013), CO₂ enhanced oil recovery (Warwick and others, 2019; Warwick and others, 2022), and helium resources (Brennan and others, 2021). The assessment of underground gas storage capacity presented here builds on the experience of these established, nationwide USGS resource assessments. Subsequently, estimates of gas storage capacity are computed at the play, basin, and regional scales and not focused on specific reservoirs or wells.

Current Underground Gas Storage Practices

Currently, about 400 natural gas storage facilities occupy underground settings and geologic settings across the United States. Starting in 2016, the Pipeline and Hazardous Materials Safety Administration (PHMSA) began inspecting and collecting data on the operations of interstate storage facilities across 31 States. There are currently about 300 of these facilities (PHMSA, 2023). State data collection on intrastate-only facilities—those not linked to interstate pipelines, which constitutes another 80 facilities, according to the PHMSA—varies widely in reporting requirements and frequency. The number of active and inactive facilities is usually tracked by regulators (PHMSA, 2023).

Storage facilities often report both working gas and base gas volumes. These reports are based on proprietary studies and measured flows and usually sum to the total gas volume. The underground storage of natural gas requires a volume of base or cushion gas that remains in the reservoir. This base gas is intended to maintain pressure support to reduce groundwater encroachment and pore space compaction in the reservoir (Katz and Tek, 1981; Plaat, 2009). The working gas volume is the gas expected to be injected into the reservoir for later withdrawal and marketplace availability, exclusive of the base gas (Plaat, 2009; U.S. Energy Information Administration [EIA], 2015). The U.S. Energy Information Administration (EIA) collects data from storage facility operators and publishes selected capacity data weekly, monthly, and annually. In 2022, based on EIA data, about 9.27 trillion cubic feet (Tcf) of total gas storage capacity, including about 4.79 Tcf of working gas storage capacity, was available in the United States (EIA, 2023b). Of the total working gas storage capacity, about 82 percent of the gas was stored in depleted hydrocarbon reservoirs, with the remainder stored in salt caverns (10 percent) and typically nonpotable aquifers (8 percent) (EIA, 2023b). Underground gas storage is currently the most evaluated and applied of all energy storage methods among the varied geologic settings used (Crotogino and others, 2018; Matos and others, 2019).

Additional Storage in Depleted Reservoirs

Depleted hydrocarbon reservoirs—primarily those that produced oil, gas, and natural gas liquids—are the most common storage setting, having been used for gas storage in the United States since 1916 (for example, near Buffalo, New York; refer to Beckman and others [1995]). Consequently, these formations are likely to continue to be used in future geologic energy storage projects. These conventional hydrocarbon reservoirs are usually well-defined by previous geophysical characterizations and development drilling and, therefore, are less costly to develop than other geologic storage options, such as saline aquifers and salt caverns (Lake and Carroll, 1986; Plaat, 2009).

Depleted hydrocarbon reservoirs have a demonstrated storage capability because they trapped hydrocarbons for as long as hundreds of millions of years before production. Regardless, the continuous monitoring of existing storage facilities is a standard industry practice (Orr and Dominguez, 2016; Callas and others, 2022). Depleted hydrocarbon reservoirs are primarily used to store natural gas, but given appropriate pressures and seals, they could be used to store liquid petroleum gas (LPG), carbon dioxide, hydrogen, ammonia (a hydrogen carrier), or compressed air. For examples, refer to Lord and others (2014), Budt and others (2016), Attanasi and Freeman (2023), and Lackey and others (2023). Natural gas can also be stored in depleted oil reservoirs, where it occupies the primary gas cap or a secondary cap formed during oil production; however, contamination of the injected gas is possible (Goodman, 1968; Plaat, 2009).

Sandstone and carbonate lithologies are both amenable to underground gas storage facilities (Wang and Economides, 2012; Schultz and Evans, 2020), and geologic considerations of existing projects are summarized in this report. Sedimentary basins, often composed of geologic formations with these lithologies could, therefore, provide substantial capacities for additional gas storage, given favorable reservoir properties (for example, porosity and permeability). Furthermore, underground gas storage projects can inject and withdraw gas from multiple zones comprising one or more geologic formations or pressure-connected intervals (also known as reservoirs).

The focus of the methodology presented in this report is assessing depleted conventional reservoirs that primarily trapped significant accumulations of gas. These reservoirs are presumed to have porosity and permeability qualities that enable holding and transmitting gas and also have working seals or caprocks that limit gas leakage.

Flow Chart and Assessment Procedure

A methodology for assessing the potential storage of natural gas (and other energy gases) and the associated capacity in depleted hydrocarbon reservoirs benefits from a robust understanding of geology, reservoir engineering, and the operational aspects of gas storage. A summary of the steps that constitute this methodology are shown in the flowchart in figure 2. The flowchart illustrates the process for (1) using databases to obtain reservoir data, (2) identifying candidate depleted reservoirs using GIS, (3) computing three probabilistic estimates for storage, and then (4) aggregating these results for further understanding. The assessment steps and inputs, along with necessary classifications, are explained in the following sections.

Step 1. Identifying Candidate Gas Reservoirs

Reservoirs are candidates for storage when depleted after primarily producing gas. Reservoir depletion is related to field abandonment, which occurs when hydrocarbon production is no longer economically favorable due to external market factors or technical conditions. These external factors can include a drop in commodity prices or a lack of affordable access to transmission infrastructure. Changing technical conditions can include a drop in reservoir pressure affecting the production drive or water production (water cut), often requiring additional disposal measures and increased expenses. Also, as described in the “Assessment Methodology Scope” section, this methodology focuses on hydrocarbon reservoirs with a GOR of 10,000 SCF/Bbl or greater. A consistent and repeatable approach is needed for identifying depleted gas reservoirs potentially available for new storage projects.

One approach for identifying depleted reservoirs is based on applying a predetermined gap in production noted at the well or entity level (for example, lease, reservoir, or field) (Varela and Buursink, 2025). If a collection of wells or producing entities associated with a specific reservoir show an extended time gap since production (greater than 5 years) or were not assigned production, then abandonment is assumed. A 5-year production gap was determined significant after a review of the full range of gaps in a commercial oil and gas production database (S&P Global Commodity Insights [SPGCI], 2023). Although production gaps ranging from months to decades were observed, wells with a 5-year or greater production gap returned to production status at a rate of less than 1 percent (Varela and Buursink, 2025). An example application of this approach is shown in figure 3, wherein both active and abandoned wells are shown around a selected reservoir under consideration in the Michigan Basin. Active wells located in and around a field under consideration and producing from the target formation indicate that the reservoir is not necessarily depleted and not a candidate for storage.

Another approach for identifying depleted reservoirs involves reviewing yearly reservoir-level production data. An example is reviewing data from commercial databases (Nehring Associates, Inc., 2017). This practice is often called decline curve analysis (Arps, 1945; Fetkovich and others, 1987; Jahediesfanjani, 2017). However, reservoir-level production data in a database are often calculated from comingled field-level production numbers, as reported by the operator. These field-level data are called unallocated production and might be allocated to reservoirs based on historical information. There is no reliable substitute for this reservoir-level information, which could complicate decline curve analysis. Secondary or tertiary recovery efforts could also alter the production record. Only reliable and allocated time-series production data can allow this approach to be consistently applied. If the reservoir shows no production for more than a fixed number of years, based on localized State-level conditions determined during the assessment, the reservoir is assumed abandoned and available for storage.

Step 2. Calculating Storage Capacity

Underground gas storage capacity can be calculated using the three governing equations defined in this step:

• the recorded production approach (eq. 1),

• the reservoir volumetric approach (eq. 2),

• or the material balance approach (eq. 3).

Each approach considers different gas and reservoir properties and includes a governing equation for calculating storage capacity, or working gas volume, specifically (Katz and Coats, 1968; Tek, 1987; Wang and Economides, 2009b). In this methodology, the three approaches are presented in order of increasing complexity; however, each can be used separately or combined using statistical approaches such as linear regression models or measurement error models to produce more accurate predictions of storage resources (Jones and others, 2024). For example, if only cumulative gas production is known, the first equation is used. If successive production and pressure values are recorded, the third equation can be used instead. As explained in the following section, the applicability of each approach relies on the availability of data about a given reservoir. A combined application of these approaches that uses input data from the Michigan Basin is shown in the “Demonstrating Storage Capacity in the Michigan Basin” section.

Material Balance Approach

For the material balance approach, which is sometimes referred to as the pressure drop approach, the storage capacity calculation is based on the difference between the depleted reservoir pressure and the initial reservoir pressure when the production of the original gas in-place started (Dake, 1978; Moghadam and others, 2011). This calculation is a net storage or working gas-capacity estimate because it is assumed that the cushion or base gas is accounted for by the volume of unproduced gas remaining in the depleted reservoir (Katz and Coats, 1968; Tek, 1987).

This governing equation (eq. 3) explains the material balance approach:

$G_{stor}=G_{prod}\bullet \left(\frac{\raisebox{1ex}{$P_{max}$}\!\left/ \!\raisebox{-1ex}{$Z_{max}$}\right.\mathrm{ }-\mathrm{ }\raisebox{1ex}{$P_{base}$}\!\left/ \!\raisebox{-1ex}{$Z_{base}$}\right.}{\raisebox{1ex}{$P_{init}$}\!\left/ \!\raisebox{-1ex}{$Z_{init}$}\right.\mathrm{ }-\mathrm{ }\raisebox{1ex}{$P_{base}$}\!\left/ \!\raisebox{-1ex}{$Z_{base}$}\right.}\right)$

,

(3)

where

G_stor

is the estimated storage or working gas capacity calculated at standard conditions, in billion cubic feet (Bcf);

G_prod

is the recorded cumulative gas production measured at standard conditions (Bcf);

P_init

is the initial (otherwise known as discovery) reservoir pressure, in pound-force per square inch absolute (psia);

P_base

is the base reservoir pressure (psia);

P_max

is the maximum storage pressure (psia); and

Z_init, Z_base, Z_max

are the pressure-respective, unitless, gas-compressibility factors.

Figure 4 is a plot illustrating the material balance or pressure drop approach, including the three pressure points and three gas volumes, respectively. The ideal plotted relation is linear when only gas pressure drive is assumed. Ultimately, this approach assumes no pressure support from outside the reservoir, such as water drive, that would make the relation nonlinear (for example, Dake [1978] and Moghadam and others [2011]). If water drive is measured, another approach may be more suitable, or the reservoir should not be assessed.

Gas storage facilities commonly operate at pressures that exceed the pressure at discovery, which is often termed “initial reservoir pressure,” to accommodate a higher working gas capacity. The extra storage pressure, known as delta pressure (P_delta), is usually set by operators at about 10 percent above the initial pressure (for example, Katz and Coats [1968] and Ibrahim and others [1970]), although this value may be adjusted as the operator learns more about the facility. To maintain the seal integrity of the storage reservoir, P_max = P_init + P_delta must not exceed the fracture pressure gradient. The fracture pressure gradient is a complex parameter that varies by depth and basin (Eaton, 1969; Osborne and Swarbrick, 1997). Therefore, during the assessment, the P_max parameter is estimated between the hydrostatic pressure and the fracture pressure based on a review of regional data for existing storage facilities or rock mechanics. As described in later sections, obtaining accurate records or estimates of initial pressure (P_init) and base pressure (P_base) are important when using the material balance approach. A gas compressibility factor (Z) may be calculated for the three pressure states to increase the accuracy of this approach. “Z” is a function of pressure, temperature, and gas composition, and is further described in the “Reservoir Hydrocarbon Properties” section below.

Step 3. Assembling Assessment Inputs and Associated Uncertainties

This section describes how the input data and related uncertainties can be assembled to calculate the most robust storage capacity. Given each approach described in Step 2, assessment inputs include datasets with gas production volumes along with various reservoir and gas properties. The following three subsections describe how these input data can be obtained. The availability of assessment data dictates the most appropriate use of one or more of the governing equations, as is demonstrated later. A demonstration of the data gathering process for two gas plays in the Michigan Basin is given in the “Demonstrating Storage Capacity in the Michigan Basin” section.

Step 4. Deriving Probabilistic Estimates and Aggregating Results

This section details probabilistically estimating and aggregating storage capacity based on a set of equations and occasionally uncertain input values. Specifically, this gas storage assessment methodology is different from other USGS energy resource assessments because three governing equations can apply. Each governing equation relies on a different approach and, therefore, requires different input parameters as described above. When inputs for all three governing equations (eqs. 1–3) are available, probabilistic outputs can be calculated using all three governing equations. The weight of each equation’s contribution is affected by the uncertainty around the respective inputs. Consequently, probabilistic estimates for storage capacity should be more representative because these consider multiple physical properties. Nevertheless, in practice, the input parameters are often less well known, and a staggered application of the approaches may be appropriate. A staggered approach is reasonable due to the varied availability of data for an assessment, and this application is demonstrated for the Michigan Basin example described in the “Demonstrating Storage Capacity in the Michigan Basin” section.

Assessment Arrangement

This section describes the spatial arrangement of the assessment, which relies on established USGS practices. The country is divided into regions and provinces, as defined by the USGS 1995 National Oil and Gas Assessment, to systematically assess the storage capacity for the entire United States (U.S. Geological Survey National Oil and Gas Resource Assessment Team, 1995; Beeman and others, 1996). While the assessment provinces are named based on major sedimentary basins, the basin boundaries do not necessarily align with published basin boundaries. Instead, the provinces are drawn to cover all areas of onshore and State waters for the United States. If typical basin boundaries were used, then areas between basins would be unassessed. Functionally, these provinces serve as geographic groups of storage reservoirs that are useful for sorting assessment input data and aggregating output estimates.

An important difference in this methodology, when compared with those used for previous USGS undiscovered oil and gas assessments (for example, refer to Schmoker and Klett [2007]) and the USGS geologic carbon dioxide sequestration assessment (Brennan and others, 2010; Blondes and others, 2013a), is that the reservoir is the basic assessment level rather than a hydrocarbon assessment unit or storage assessment unit (porous interval along with a regional seal), respectively. This assessment level is further prescribed because reservoir-specific data are available in important databases described in the “Available Data Sources” section below.

Reservoirs within each province are grouped into plays for this assessment. Plays are defined as “a set of hydrocarbon accumulations sharing similar geologic, geographic, and temporal properties, such as source rock, migration pathway, timing, trapping mechanism, and hydrocarbon type” (Magoon and Schmoker, 2000). The EIA (2024) similarly defines a play as a “set of known or postulated oil and gas accumulations sharing similar geologic, geographic, and temporal properties, such as source rock, migration pathway, timing, trapping mechanism, and hydrocarbon type.” Besides providing an additional assessment structure, organizing at the play level helps quantify assessment inputs, especially when reservoir-specific properties are unknown. In effect, plays become useful analogs, as they include geologic information in their descriptions. The application of reservoir- and play-level data is further explained in the following section.

Available Data Sources

This section describes several potential data sources for the assessment. Usually, a combination of proprietary commercial and public domain databases are used by geologists to gather input data and related uncertainties. Proprietary data sources currently (2025) available at the USGS include the Nehring Associates, Inc. (2017) “Significant Oil and Gas Fields of the United States Database” and the “Comprehensive Resource Database” (CRD) (Carolus and others, 2018; Warwick and others, 2019). As a primary source, the CRD provides hydrocarbon production data, and reservoir fluid and gas properties derived from the legacy Nehring Associates, Inc. (2017) database along with modern production and drilling data, by well, from IHS Inc. (2012).

Even though the CRD used recent production volumes from IHS Inc., the assessment geologists may refer to the S&P Global Commodity Insights (SPGCI), ENERDEQ database (SPGCI, 2023) for more current production information as necessary. Note that well- and lease-level production data from SPGCI were combined to field level in the CRD and that for fields where the two databases matched, the updated production data for SPGCI were allocated to reservoirs in the Nehring database according to each reservoir’s historical shares (Nehring Associates, Inc., 2017; Carolus and others, 2018; Warwick and others, 2019; SPGCI, 202393).

Lastly, the Nehring database only considers significant fields with a known recovery (the sum of cumulative production and proved reserves) of 500,000 barrels of oil equivalent or 3 billion cubic feet of gas-equivalent barrels, or more (Nehring Associates, Inc., 2017). Therefore, this assessment focuses on calculating the potential gas storage capacity for larger reservoirs that can more effectively add to a storage capacity assessment.

When available, other publicly available reservoir production and property databases can be used to verify and supplement the estimates and ranges of reservoir data found in the CRD. Potential ancillary sources include the Department of Energy-contracted Gas Information System (GASIS) database (Hugman and others, 2000), the Tertiary Oil Recovery Information System (TORIS) database originally developed by the National Petroleum Council and rereleased by the National Energy Technology Laboratory (NETL) (Guinn and Remson, 1999), and supplementary data compiled by the Appalachian Oil and Natural Gas Research Consortium (1996), and others. Commercial databases, like the CRD and SPGCI described above, provide consistent national data otherwise too complicated to gather and quality control using public (often State-level) databases. Ultimately, database entries are verified as valid inputs by the assessment geologists, who typically review complementary play- or basin-specific literature that could help support these estimates.

As described earlier, when reservoir-specific data are unavailable, play or province averages may be computed in the CRD (Carolus and others, 2018; Warwick and others, 2019). More specifically, when reservoir pressure or temperature data are unavailable, basin-averaged gradients are used, although these do not necessarily account for local variations (for example, refer to Bahr and others [1994] and Buursink [2014]). A textbook hydrostatic gradient for saline produced water (0.45 psi/ft) may be assumed where initial pressure data do not exist. Sometimes, input data are so sparse that a national average is needed, as with initial gas saturation (S_gi) (fig. 5). Figure 5 shows the distribution for S_gi values used in the assessment as 0.50 for the minimum, 0.67 for the most likely, and 0.84 for the maximum.

For the other inputs, the assessment geologists, through expert elicitation, assign minimum, most likely, and maximum estimates for the input parameters consistent with reservoirs in the given geologic play. As is common in USGS resource assessments, a Beta-PERT distribution is assumed (fig. 5), which is a three-parameter special case of the four-parameter Beta distribution and is more amenable to geologic data (for example, Vose, [1996] and Olea [2011]). Regardless, reservoirs in the CRD are listed by the geologic plays and petroleum provinces identified during the 1995 USGS National Oil and Gas Assessment (NOGA) project (U.S. Geological Survey National Oil and Gas Resource Assessment Team, 1995; Beeman and others, 1996). This play and province arrangement, as described earlier, provides a convenient organizational and geographic structure through which the previously described assessment can be conducted.

Assessment Scope and Storage Correction Factor

This section elaborates on the scope and correction factor applied in this assessment (eq. 2). The consideration of reservoir property changes due to pressure depletion (resulting in, for example, reservoir compaction or induced seismicity) that typically occur during end-stage hydrocarbon production or withdrawal (for example, Chang and others, [2014] and Van Wees and others [2014]) may be within the scope of this assessment. Reservoir property changes related to gas leakage and contamination are not within the scope of this assessment.

Characterizing individual reservoirs for leakage risk—for example, stratigraphic or fault seal integrity—is beyond the scope of this assessment. This time-consuming work requires core plug analysis or seismic data interpretation (as in Knipe and others [1997] and Foschi and van Rensbergen [2022]) but is recommended when storage facilities are developed. Hydrocarbon reservoir risks should be considered broadly (for example, Roberts and Cordell [1980] and Dandekar [2006]); however, if an abandoned reservoir has trapped gas before being produced, it is assumed suitable for future gas storage (as demonstrated by existing underground storage facilities described in the earlier “Current Underground Gas Storage Practices” section). Another recommendation pertains to assessing gas deliverability when evaluating future gas storage facilities, such as conducting reservoir permeability or flow tests, and modeling well spacing (refer to Tek [1987] and Katz and Lee [1990]). Consideration of microbial activity that can contribute to contamination of the stored gas in underground storage facilities such as methanotrophy and sulfate reduction (as in Hemme and van Berk [2017] and Molíková and others [2022]) is beyond the scope of this assessment.

A storage correction factor may be determined during the assessment when supporting reservoir data and observations from the field operator are available. This factor can quantify and lump reservoir issues that might arise during injection or withdrawal operations at a storage facility. These storage issues may include (1) groundwater intrusion, (2) formation compaction, or (3) changes in storage efficiency, among others. Groundwater intrusion can occur when water infiltrates the pore space originally used for gas storage (for example, refer to Katz and Coats, [1968] and Wang and Economides [2009a]). Reservoir compaction can occur when formation pore space is lost due to overburden pressure reducing the porosity or permeability, particularly in reservoirs hosted in young sedimentary rocks before significant lithification occurs (for example, Martin and Serdengecti, [1984] and Zoback and Smit [2023]). Storage efficiency is the fraction of the total available pore space accessible for gas injection (as in Barbieri and others [2011] and Al-Shafi and others [2023]). Decreased storage efficiency or reduced access to pore space (as with clogging) can occur due to geochemical or microbial processes (refer to Eddaoui and others [2021] and Azin and Izadpanahi [2022]). These reservoir issues can usually be quantified as unitless slope and intercept variables in the recorded production approach equation and by the storage utilization factor in the volumetric approach (Katz and Coats, 1968; Jones and others, 2024).

Motivation for Demonstration Evaluation

This section presents an example of a storage capacity estimate by (1) describing the province geology and two hydrocarbon plays of the Michigan Basin (2) and then demonstrating a probabilistic calculation of natural gas storage capacity. Steps discussed in this section refer to Steps 1–4 defined in the “Assessing Storage Capacity” section. Initially, candidate reservoirs are identified and assessment inputs are determined. Subsequently, assessment output distributions are evaluated and total storage capacity results are aggregated. This section provides example guidance for consistent USGS assessments of hydrocarbon provinces across the United States.

Michigan Basin Province

Reservoirs in the Michigan Basin Province are used as examples to demonstrate the storage capacity methods presented in this report. This province was selected because of the significant number of existing underground gas storage facilities and the reliable reporting of capacity in available reservoirs (Michigan Public Service Commission, 2024). According to the EIA (2023b), Michigan has over 1 Tcf of total storage and 690 Bcf of working storage primarily available in depleted reservoirs. Because reliable data for a large number of reservoirs were available, Jones and others (2024) were able to test the assessment equations described in the “Step 2. Calculating Storage Capacity” section. This example workflow on reservoirs in the Michigan Basin is expected to be applicable for assessing other U.S. petroleum provinces, although adjustments due to reservoir presence and data availability are expected.

Most of the Michigan Basin is located in the State of Michigan and is represented by the USGS assessment province of the same name (Gautier and others, 1996). The Michigan Basin Province includes the Lower Peninsula and eastern Upper Peninsula of the State of Michigan, a large part of eastern Wisconsin, 12 counties in northeast Indiana, 5 counties in northwest Ohio, and adjoining parts of the Great Lakes. Almost the entire assessment province (about 119,000 square miles) is underlain by sedimentary rocks of the Michigan Basin in the United States (excluding only small parts of the basin in Canada) (Dolton, 1996; Swezey and others, 2015).

The Michigan Basin, with its roughly circular shape, is a classic interior cratonic basin, having boundaries defined by a series of bedrock highs and structural features. Clockwise from the west, these highs and features are the Wisconsin arch, the contact with Precambrian Canadian Shield, the Algonquin arch, the Findlay arch, the Logansport fault, the Kankakee arch, and the Sandwich fault (Catacosinos and others, 1990; Dolton, 1996; Swezey and others, 2015). Basin evolution, structure, and stratigraphy are summarized by Fisher and others (1988), Catacosinos and others (2001), and more; in focusing on process steps and not framework geology, these publications give further information on basin physiography and detailed formation mapping and figures.

Oil and gas were first discovered in the Michigan Basin at the Port Huron field in 1886, whereas modern hydrocarbon production in the basin began in 1925 with the discovery of the Saginaw field (Catacosinos and others, 1990). Oil and gas in the Michigan Basin are produced from reservoirs from the Middle Ordovician to the Pleistocene (Dolton, 1996). A combination of stratigraphy and structure allows a division of the Michigan Basin fields and prospects into 16 principal hydrocarbon plays (Dolton, 1996), consistent with the definition of a play described in the “Assessment Arrangement” section.

Geology of Demonstration Plays

This section describes the petroleum geology for two example hydrocarbon plays in the Michigan Basin along with the consideration of other plays in the province. Based on the number of existing facilities and the potential for additional storage capacity, the contrasting Mississippian Sandstone Gas and Clinton Structural Plays are considered representative settings for demonstrating this methodology. The Mississippian Sandstone Gas Play consists of small, shallow Mississippian sandstone reservoirs on anticlinal structures in the central part of the Michigan Basin. The reservoirs are part of a primarily clastic sequence that shows considerable stratigraphic variability. Most of the gas production is from the informal stray sandstone of the Michigan Formation and from the Marshall Sandstone (Dolton, 1996); Dolton assigned non-gas-producing Devonian Berea Sandstone reservoirs to a separate play. Representative play reservoir depths range around 1,500 ft; net reservoir (as opposed to gross) thicknesses range between 5 and 20 ft; and porosities range between 10 and 15 percent (Dolton, 1996; Nehring Associates, Inc., 2017). Catacosinos and others (1990, p. 574) and Swezey and others (2015, p. 54) provide additional details and maps for these reservoir intervals.

The Clinton Structural Play comprises gas reservoirs in Silurian Clinton Formation carbonate rock and trapped in deep structural features in the northern Michigan Basin. Most of the reservoirs in this play have been deep pool additions at fields with previously discovered shallower reservoirs. Reservoirs are typically open-shelf limestones that have dissolution-generated porosity and have undergone dolomitization (Dolton, 1996). Equivalent Silurian Burnt Bluff Group dolomite gas reservoirs were also considered in this play interval (Swezey and others, 2015). Representative reservoir depths range around 7,000 ft, net reservoir thicknesses range from 20 to 100 ft, and porosities are around 5 percent (Dolton, 1996; Nehring Associates, Inc., 2017). Catacosinos and others (1990, p. 590) and Swezey and others (2015, p. 128) provide additional details and maps for these reservoir intervals.

This demonstration does not include results from the large Northern and Southern Niagaran Reef Plays (Dolton, 1996). These Michigan Basin plays already host significant volumes of gas storage, particularly the northern trend, which is well characterized (Haagsma and others, 2020; Jones and others, 2024). These formations are planned for assessment in the future and may serve to evaluate storage in settings that include vuggy porosity (due to the dolomitized reef structure of the reservoirs) akin to cavern storage. The deeper reservoirs in the Ordovician Sandstone Gas Play in the Michigan Basin (Dolton, 1996) would also be evaluated as part of planned future work, although some may be too deep for conventional use given the assessment-depth cutoff of 8,000 ft, as explained in the “Gas Storage Reservoir Classes” section.

Demonstration Assessment Approaches

As part of Step 1 in this demonstration in the Michigan Basin, 2 available reservoirs were identified out of 16 significant fields in the Mississippian Sandstone Gas Play, and 5 available reservoirs were identified out of 8 significant fields in the Clinton Structural Play. These depleted reservoirs were identified using the Varela and Buursink (2025) approach described earlier, which is based on well abandonment. An example of an available Michigan sandstone reservoir in the Evart field surrounded by nonamenable fields is shown in figure 3. Figure 3 shows selected fields in the Michigan Basin with corresponding active and abandoned wells. As is the case in multiple Michigan Basin Province plays (fig. 3), many reservoirs are already used for gas storage (MPSC, 2024). These reservoirs are considered unavailable and are not assessed. Fields not meeting the significant field cutoff from Nehring Associates, Inc. (2017) are also ignored.

In preparing to calculate storage capacity, values for deterministic and probabilistic inputs for the demonstration reservoirs were from the data sources described in the “Available Data Sources” section. The following input data values were gathered: cumulative gas production, initial gas saturation, gas formation volume factor and gravity, reservoir thickness and area, porosity of the net interval, and discovery and depletion pressures. Table 1 shows the representative input ranges, including the maximums and minimums for the two plays considered in the Michigan Basin (one is generally deeper than the other). While this table does not show proprietary data at the reservoir level, it does show the differences in the reservoir properties between the two plays. Also, when reservoir-specific inputs were unavailable, play- or basin-level, or even national averages, were used as described in the “Available Data Sources” section (as with S_gi).

Given the available inputs for the Michigan Basin plays described above, storage capacities were calculated using all the approaches detailed in Step 2. For this demonstration, staggered calculations relied on recorded cumulative production values along with the reservoir volume to derive a depletion pressure (P_base) that was then used in the material balance equation to calculate storage capacity. In effect, the more robust material balance approach was used, while depletion pressure was derived using gas production records and reservoir volume calculations according to Carolus and others (2018). This staggered approach may be appropriate for assessing remaining plays and provinces when not all input data are available as described in Step 3.

Example Michigan Basin Probabilistic Results

The results of the capacity estimates are given as histograms from Monte Carlo simulations in which each input distribution was sampled 10,000 times. This is part of Step 4 and was coded in R (R Core Team, 2024). Example probabilistic output distributions are provided in figure 6 for the seven reservoirs identified as available for storage in the two plays in the Michigan Basin. Table 2 lists the P₅, P₅₀, and P₉₅ probability percentiles that represent the 5-, 50-, and 95-percent probabilities, respectively, that the true storage capacity is less than the value shown. The terminology used in this report follows standard statistical practice (and may differ from that used by the petroleum industry [Everitt and Skrondal, 2010]) such that percentiles, or quantiles, represent the value of a variable below which a certain proportion of observations fall. Table 2 provides the anonymized entity identification and reservoir depth class.

Also listed in table 2 are example aggregated results for the total storage capacity of the candidate reservoirs in each of the two example plays in the Michigan Basin. Additional gas capacity, ranging from 47 to 64 Bcf, with a likely capacity of 54 Bcf, was estimated, representing an 8-percent increase in the existing total storage capability in Michigan. The capacity percentile calculations used the aggregation method described in Step 4. The correlation coefficients for the aggregation at the reservoir level (ranging from 0.82 to 0.86) are primarily based on reservoir porosity and original gas saturation. Even though not all plays in the Michigan Basin were assessed for this example, as a demonstration, the play level results were aggregated at the basin level using a correlation coefficient (0.6) that primarily considered the different lithologies, depth ranges, and average porosities of the two plays (table 2).