Abstract

Geothermal resources have the potential to fulfill a significant portion of the low-temperature (30–100 °C) thermal energy demand in the United States. Investment risk at the exploration stage is a primary factor limiting the development of geothermal energy projects, due to the high cost of drilling and limited reservoir data. An approach to reduce this risk is to target proven, well-characterized conventional oil and gas reservoirs. We examined the suitability of the Trenton–Black River gas fields of southern New York as geothermal reservoirs. These highly productive hydrothermal dolomite fields occur within long, narrow normal-fault–bounded, en echelon grabens that are scattered with saddle dolomite-lined vugs, fractures, and breccia. The Quackenbush Hill field was analyzed using existing data sets with geothermal purposes in mind. Key geothermal reservoir characteristics examined here include rock temperature, porosity and permeability, stimulation potential, and the risk of inducing seismicity. Results indicate that the Quackenbush Hill field would produce temperatures of ∼91 °C from a dolomite reservoir with sufficient average horizontal permeability, low vertical permeability, and significant vertical and horizontal anisotropy. In the case that adequate flow rates cannot be achieved in practice, stimulation is a feasible option from the perspective of well-field design for optimal heat sweep; however, higher-resolution data are necessary to constrain the risk of inducing seismicity. We demonstrate the technical feasibility of transitioning conventional gas fields into geothermal heat-producing reservoirs, setting the stage for future consideration of the economics of a petroleum-to-geothermal transition.

INTRODUCTION

Sustainable and affordable sources of clean energy are widely sought after in the United States and around the world as a way to reduce dependence on fossil fuels. Geothermal energy has been commonly suggested as a part of the solution for renewable baseload electricity needs in the United States (e.g., Tester et al., 2012), but few places on Earth are host to the required temperatures for electricity generation at depths that allow a project to be economically feasible. An additional challenge faced by the geothermal industry is adequate subsurface reservoir data prior to the expensive drilling of geothermal wells. We suggest that repurposing depleted conventional oil and gas fields for low-temperature (30–100 °C) geothermal heat production is a logical, near-term solution to address the geographical and data limitations of geothermal energy. Harnessing geothermal energy for direct-use heat instead of electricity expands geographical opportunities into regions with lower geothermal gradients, significantly reducing the required depth of drilling, associated capital costs, and exploration risk. Furthermore, many sedimentary basins have already been explored and drilled extensively by the oil and gas industry and thus provide available reservoir data and infrastructure that may be reused to harness otherwise wasted heat remaining in depleted hydrocarbon reservoirs.

Heat utilized at temperatures between 30 and 100 °C represents over half of the total thermal energy consumption in the United States, including applications such as space and water heating, food drying, industrial processes, animal husbandry, refrigeration, or greenhouses (Lindal, 1973; Fox et al., 2011). Therefore, geothermal energy heat production is a logical choice to fill the high demand for low-temperature energy applications. In a typical closed-loop low-temperature geothermal system, relatively cool water is injected into one or more deep wells, circulated through a permeable body of hot rock, and pumped to the surface from another well. The geothermally heated water is passed through a series of heat exchangers in a heating plant and reintroduced back into the reservoir to be reheated (Fig. 1).

We explore a gas play in the northern Appalachian Basin of the eastern United States, where hydrocarbon exploration and drilling has been ongoing for 150 years. As shown by recently published geothermal maps for the United States based on bottom-hole temperature (BHT) data (Blackwell et al., 2010; Shope et al., 2012; Stutz et al., 2012), average geothermal gradients in the Appalachian Basin range from 20 to 25 °C/km, equating to required well depths of 1.5–5 km for direct use, depending on the application. Animation 1 shows that temperatures in the basin range between 50 and 150 °C at depths of 3.5 km to 4.5 km (Blackwell et al., 2011). Though hydrocarbons have been produced from the Appalachian Basin for a century and a half, most of the oil and gas plays are relatively tight or have natural fracture systems with low permeability (Roen and Walker, 1996). Where there are exceptions with higher permeability, there is a much lower risk for geothermal energy exploitation. The Trenton–Black River (T-BR) hydrothermal dolomite (HTD) gas fields of southern New York are considered one of those exceptions.

The T-BR gas fields of southern New York (Fig. 2) were discovered in 1986 and have been producing large volumes of natural gas since 2000. These fields have previously been characterized as having low-matrix porosity but moderate to high permeability, due to the presence of vugs (small cavities in the rock typically lined with saddle dolomite), brecciated rock, and multiple fracture sets (Smith et al., 2009). The fields vary in size but are generally long and narrow normal-fault–bounded grabens, ranging from 5 to 20 km in length and 0.5–3 km in width. Studies have shown that these dolomite fields are hydrothermal in origin, having formed via hydrothermal alteration of limestone by hot saline magnesium-rich fluids that traveled upward along faults and concentrated below the contact between the Trenton and Black River Formations (Davies and Smith, 2006; Smith et al., 2009).

Stratigraphic and Structural Setting

After the Grenville orogeny 1.1 billion years ago, the supercontinent in which North America was embedded experienced a long period of rifting that left an imprint of failed rift zones, extensional faults, and strike-slip faults (Thomas, 1991; van Staal, 2005). The region comprising the state of New York was part of a passive margin from the Cambrian to Late Ordovician, which allowed for the deposition of thick strata on the rifted basement. Carbonates of the Black River Formation are muddy and fine grained, including mudstone, wackestone, and packstone, indicative of deposition on a shallow tropical carbonate ramp (Smith et al., 2009). Carbonates of the Trenton Formation were deposited conformably above the Black River and include deep-water argillaceous limestone and high-energy, shallow-water packstone (Brett and Baird, 2002). Tectonic activity resumed when the Taconic orogeny began in the Late Ordovician, concurrent with the end of the deposition of the Black River and Trenton Formations.

Previous studies agree that the fields are negative flower structures acting as en echelon Riedel shears associated with a large-scale, northwest-trending, left-lateral transtensional basement-rooted wrench fault that experienced episodic reactivations (Hurley and Budros, 1990; Rasmussen et al., 2003; Davies and Smith, 2006; Smith et al., 2009; Slater and Smith, 2012). The timing of this deformation is not well constrained, but because the graben faults do not extend upwards past the Trenton Formation (Smith et al., 2009), deformation and hydrothermal alteration must have taken place soon after deposition of the Trenton Formation but before the Utica Shale was deposited. According to Rasmussen et al. (2003), hydrocarbon-stable, magnesium-rich paleofluids that formed in the basement by the serpentinization of peridotite contemporaneous with the Taconic orogeny, flowed upward along the wrench fault system and through the Riedel shears causing dolomitization.

Dolomitization in these reservoirs had a strong structural control and a limited stratigraphic control (Marner et al., 2008; Smith et al., 2009). As shown in Figure 3, dolomite is found in all facies of the Black River Formation, where it is concentrated along faults and fractures, most commonly along the hanging wall of subvertical, ENE-trending normal faults (Davies and Smith, 2006). Dolomite is occasionally found within the Trenton Formation. Outside the grabens, the unaltered limestones of the Black River and Trenton Formations are impermeable. The dolomite displays increased lateral continuity in the uppermost portion of the Black River Formation, where hydrothermal fluids are thought to have accumulated and percolated horizontally within the graben. Hydrocarbon accumulation occurred following the Late Ordovician dolomitization of the Black River limestone, all of which took place within the structural confines of the fractured grabens.

Study Area: Quackenbush Hill Field

The Quackenbush Hill field (Figs. 24) was chosen for a case study analysis to determine the potential to use these HTD fields as geothermal reservoirs. Quackenbush Hill field comprises two ENE-trending, en echelon fault-bounded grabens, and measures ∼13 km in length and 3 km at its widest point (Marner et al., 2008). The field consists of a total of 14 wells, including the most directional wells of any T-BR field in the state. Quackenbush Hill field is of particular interest because it is one of the highest gas-producing Trenton–Black River fields in New York, it is host to the best producing well in the play (Smith et al., 2009), and it is within 15 km from the population and business centers of Corning and Elmira, New York.

We evaluated the Quackenbush Hill field as a potential geothermal reservoir considering the following criteria: thermal availability, reservoir quality, potential for stimulation, and risk of induced seismicity. Among the necessary conditions for extracting a commercially feasible amount of heat from a rock reservoir, several relate to fluid flow through the rock. Effective heat transfer from the rock into the fluid, referred to as heat sweep, requires reservoir permeability high enough to transfer the quantity of heat to meet the project energy goals. For geothermal applications, suitable reservoir permeability is generally considered to be in the range of hundreds to thousands of millidarcies (mD) or greater. Consequently, we focused on the identification of data sets that pertain to thermal availability and permeability in the reservoir.

Much of the existing data from the Quackenbush Hill field is proprietary. Two-dimensional seismic data have been collected in the past but were not available for this analysis. Two vertical cores and one horizontal core were drilled in the Quackenbush Hill field; aside from those data population statistics, the raw data are not publicly available. The first vertical core, Gregory #1446, which was drilled outside the graben wall (Fig. 3), sampled both the Trenton and the Black River Formations but only penetrated tight limestone. The Gregory #1446A sidetrack core (Fig. 3), also vertical and 64 m from the first, penetrated the interior of the graben and sampled the same stratigraphy as #1446, but its rocks are completely dolomitized, with an average porosity of 2.7% and an average permeability of 22.5 mD (Marner et al., 2008). Data from these two vertical cores exhibit a permeability range of 0.01–500 mD (Marner et al., 2008; R. Jacobi, 2014, personal commun.). The horizontal core, Schwingle 2 Hz, was bored perpendicular to the trend of the field and sampled 10 m of dolomite within the graben (Fig. 3). It has an average porosity of 3.2%, an average permeability of 0.11 mD, and shows numerous vugs, occasional veins with void space, horizontal and vertical stylolites (some of which are open), and multiple generations of fractures (Marner et al., 2008).

In the case that Quackenbush Hill field meets the necessary criteria to be repurposed for geothermal heat extraction, it is likely that the other T-BR fields are good candidates as well; although the fields in the T-BR play vary in size and total gas production, they are similar in origin and structure. Our results are applicable to both the geothermal and petroleum industries, in addition to municipalities and governmental bodies. Similar analyses on other conventional oil or gas fields that are ramping down production may guide future decisions about the transition from hydrocarbon production to geothermal heat production.

METHODS

Reservoir Quality

Without access to field-specific permeability data, we looked outside the Quackenbush Hill field at neighboring producing fields in search of patterns in reservoir quality that could be applied to the Quackenbush analysis. The Whiteman #1 core, which sampled the Black River Formation in a nearby reservoir, was analyzed by CoreLab, Inc., for air permeability (maximum, 90° to the direction of maximum, and vertical), helium porosity, and bulk density at a confining stress of 2.7 MPa (New York State Museum, 2014). Measurements were taken at variable intervals, from 0.1 to 1.2 m. Because we are interested in the permeability of water through the reservoir, not air, the permeability data were corrected for the Klinkenberg effect using the power-law correlation derived for carbonates in Al-Jabri et al. (2015), where kw is the permeability of the rock with water, and kg is the permeability of the rock with gas, both in units of millidarcies (mD): 
graphic
Because the core permeability data are sparse but wireline-log–porosity data are abundant and available in a state database, we derived from core porosity and permeability data an empirical relationship to apply to well-log–derived porosity from nearby gas fields, and thereby extended the reservoir analysis. We chose to apply a power-law fit to the core data, based on goodness of fit and because power-law fits are most realistic for carbonates (Jennings and Lucia, 2003). Porosity data were acquired from neutron porosity hydrogen index (NPHI) well logs available for download on the Empire State Oil and Gas Information System database (ESOGIS, 2014). Ten of the fourteen Quackenbush Hill wells possess NPHI logs recorded every 15 cm in the gas-producing zone of the Black River Formation, for a total of 15,760 NPHI recordings. Machine-error outliers were removed using median absolute deviation, leaving 8380 valid recordings. Dolomite was distinguished from limestone using the photoelectric factor log, in which dolomite responds as ≤3.14. Apparent neutron porosity was adjusted where the matrix was dolomite, using equivalence charts for the appropriate wireline logging tool, which in this case was a Schlumberger compensated neutron logging (CNL) tool for all wells in the field (Schlumberger, 2009). Where density porosity exceeded apparent neutron porosity, the gas excavation effect was corrected to true porosity using the iterative process described by Bassiouni (1994). The presence of gas in a reservoir underestimates the neutron porosity reading because the CNL tool, which responds to hydrogen content in the formation, is calibrated to read porosity in a reservoir saturated with fresh water. For the remainder of the readings that were not affected by gas, true porosity was calculated using: 
graphic
where ϕ is true porosity, ϕN is apparent neutron porosity, ρma is matrix density, and ρb is bulk density log reading (Bassiouni, 1994). We assumed a matrix density of 2.85 g/cm3 for dolomite and 2.71 g/cm3 for limestone, based on bulk density measurements from the Whiteman #1 core.

Stimulation

If permeability is not high enough at Quackenbush to sustain high flow rates, a design option to consider is stimulation by hydraulic shearing (Mode II or III) of preexisting fractures in the reservoir (Cladouhos et al., 2014; Rinaldi et al., 2014). Mode II and III fractures experience slip parallel to the plane of the fracture, rather than outward displacement perpendicular to the plane of the fracture, which produces Mode I fractures.

Because stress-field data specific to the Quackenbush Hill field do not exist, we assume stress-magnitude gradients from a nearby data set. Stress orientations were estimated by extrapolating between two nearby locations that have available data. The first data set, which includes stress orientations and magnitudes, comes from the Auburn geothermal well in Auburn, New York (Fig. 2), ∼90 km northeast of Quackenbush Hill field, where the top of the Black River Formation occurs ∼1730 m shallower than in the Quackenbush Hill field. By means of mini-hydraulic fracturing tests and borehole televiewer surveys, Hickman et al. (1985) determined that the minimum horizontal stress, SH,min, increases linearly with depth and that the maximum horizontal stress, SH,max, increases approximately linearly in an irregular fashion and is oriented approximately N83°E ± 15°. Our study used a linear fit of this data set to determine the maximum and minimum horizontal stress gradients, which were then used to calculate the horizontal stress magnitudes at a depth of 3 km, the average depth of reservoir production at Quackenbush Hill field. SV, the lithostatic load, was computed using an average density of 2600 kg/m3 for sedimentary rocks through the entire sedimentary column (Manger, 1963).

Our second source of stress orientations comes from 20 km south of Quackenbush, where natural gas production relies on high-pressure hydraulic fracturing of Mode I fractures to stimulate shale-gas production. Since Mode I hydraulic fractures propagate parallel to the direction of maximum horizontal stress (Hickman et al., 1985), the orientation of SH,max (N68°E) can be inferred from the orientation of the horizontal legs of wellbores. Stress magnitudes were not determinable at this location.

Subsequently, an approximate orientation of the principal stresses at Quackenbush Hill field was determined using a linear extrapolation of the azimuth of SH,max from Auburn, New York (Hickman et al., 1985) to the location of the shale gas production wells in northern Pennsylvania (Fig. 2). To check the accuracy of this extrapolation, we compared our result for SH,min with the instantaneous shut-in pressures (ISIP) recorded from Quackenbush Hill field wells (IHS U.S. Well Data, 2013) and assigned error ranges based on the deviation of the ISIP values from our estimated SH,min.

By applying our stress-field data to the principles of Mohr’s circle of stress and Byerlee’s law (Byerlee, 1978), we calculated the minimum pore-fluid pressure required to hydroshear existing fractures, and the orientation of such fractures. The injected fluid pressure (Pf*) required to hydroshear existing fractures is given by: 
graphic
where σ1 and σ3 are the maximum and minimum horizontal stresses, respectively, in this stress regime, and μ is the sliding coefficient of fractured rock.

Induced Seismicity

Induced seismicity occurs when high-pressure injection of fluids changes the pore pressure in the subsurface and inadvertently causes slip on large faults (Talwani and Acree, 1984). At Quackenbush Hill field, the two sets of subvertical master faults that bound the grabens (5–8 km in length; Fig. 4) are most worrisome for reactivation given their relatively large surface areas and proximity to the aforementioned urban centers. The risk of inducing seismicity must be assessed whether or not stimulation is a feasible option. To calculate the tendency for slip along these faults, we performed a slip-tendency analysis (Morris et al., 1996) by comparing the sliding friction coefficient (μ) with the ratio of the shear stress (τ) and effective normal stress (σn*) on the faults. That ratio is given as: 
graphic
where θ is the angle between σ1 and the pole to the faults of interest, and pf is the pore-fluid pressure. The faults are oriented subvertically, approximately dipping 85° (Jacobi, 2003; Rasmussen et al., 2003) and strike between N75°–80°E, as estimated from map-view depictions of the fields (Smith et al., 2009; Slater and Smith, 2012), well locations, and from the orientations of lineaments detected by Landsat imagery (Earth Satellite Corporation, 1997; Jacobi, 2003).

Thermal Resource

In order to calculate the available and recoverable energy within the Quackenbush Hill field, an estimate of the reservoir temperature is needed. Eight wells targeting gas in the Quackenbush Hill field have recorded bottom-hole temperatures (BHT) in the Black River Formation. BHT data often require a temperature correction to account for the influence of cooler drilling fluids interacting with the rock. Whealton et al. (2015) developed a local BHT correction that varies depending on the drilling fluid. The following correction was applied to the Quackenbush Hill field wells, which were drilled with a polymer-based fluid: 
graphic
where BHTc is the corrected BHT in °C, BHTo is the original BHT in °C, Xm is the percent fraction of mud or polymer in the drilling fluid (in this case, Xm = 1), and d is depth of the BHT recording in meters. Temperatures at the depth of the reservoir were back-calculated from the corrected BHT data using the thermal model developed by Smith et al. (2015) and the depth to the top of dolomite determined from well logs. From those eight data points, an average reservoir temperature for Quackenbush Hill field was used to calculate its total reservoir energy.
We define total reservoir energy, qtot, as the maximum thermal energy in place within a certain volume of rock. The total reservoir energy in place for the Quackenbush Hill field is estimated using the relationship: 
graphic
where ρr and ρw are the densities of the rock and water in kg/m3, V is the volume of the reservoir in m3, ϕ is the porosity of the rock, Cp,r and Cp,w are the heat capacities of the rock and water in kJ/kg/°C, and ΔT is the temperature differential between the reservoir and the water reinjection temperature (Tester et al., 2012). We approximate a reservoir area of 13 km by 3 km (Fig. 3), based on the geographic extent of the gas-producing wells and what is known of the width of the two Quackenbush grabens from seismic data. The gas-producing interval of the reservoir, 33 m, was used as a conservative estimate of reservoir thickness.
All of the thermal energy in place cannot be extracted from the reservoir due to limitations of the reservoir geometry and inherent heat losses in the surface operations. What matters to the economics of the geothermal heating application is the recoverable thermal capacity of the reservoir (qrec), which is estimated using the relationship: 
graphic

Here, t is the lifetime of the heating system in seconds, R is the recovery factor used to estimate the amount of thermal energy that can physically be extracted from the reservoir, and η is the surface efficiency, which accounts for heat loss during distribution. For this calculation, we assumed a typical geothermal project lifetime of 30 years. The recovery factor largely depends on the amount of surface area available between the water and rock, which is dependent on reservoir heterogeneity and the fluid-flow regime (i.e., porous media, fractures, vugs, or any combination). In this analysis, we assumed a range of recovery factors, from conservative to optimistic, based on the results of the reservoir analysis, discussed below. The surface efficiency scales with distribution distance and piping insulation (0.1–1 °C/km for insulated pipes) (Ryan, 1981). Surface efficiency was approximated using the maximum distance from Quackenbush Hill field to Elmira (15 km) and an average temperature loss for insulated pipes (0.5 °C/km). Fluid losses were neglected for this calculation. The resulting value was applied to a hypothetical design of a geothermal district heating system to estimate the number of homes that could be heated.

RESULTS

Reservoir Quality

Klinkenberg-corrected permeability data (maximum, 90°, and vertical) are presented in Figure 5, plotted against measured core porosity. Though the spread is very large, the maximum and 90° permeability data were fit together by a power-law model with an R2 of 0.52. That fit is: 
graphic
where kL is Klinkenberg permeability, and ϕ is porosity. Vertical permeability was neglected from this fit because average vertical permeability (2.6 mD) is three orders of magnitude lower than both the average 90° and average maximum permeability, and was deemed to be negligible and not of reservoir quality. Average core porosity is 7%, average maximum permeability is 4680 mD, and average 90° permeability is 2100 mD (Table 1).

When the log of maximum permeability is plotted against porosity categorized based on core features (i.e., vugs, fractures, etc.), there is a distinguishable grouping pattern (Fig. 6). Fractures that are either isolated or those adjacent to small vugs do not exhibit porosity exceeding 4%. Nevertheless, fracture permeability spans six orders of magnitude, and exceeds 100 mD where co-located with vugs. Porosity associated with vugs increases with vug size, reaching nearly 27% where large vugs are co-located with small and medium vugs. Permeability associated with vugs spans four orders of magnitude, from 0.5 mD to 10,240 mD.

The power-law model from the core was applied to the corrected NPHI porosity data from ten wells. The maximum porosity recorded in the reservoir is 53%, and the minimum is 0.002%. Applying Equation 8 to the NPHI data, the average calculated permeability in Quackenbush Hill field is 120 mD. This porosity-based method predicts a maximum permeability in the reservoir of ∼36,000 mD, which is within the same order of magnitude as the highest permeability detected in the core.

Calculated permeability from log porosity was plotted with depth (Fig. 7), corrected to account for the difference between measured borehole length and total vertical depth. The results show three bands of very high porosity at depths of 2850, 2950, and 3020 m; these bands show porosity up to 32%, 53%, and 19%, respectively. Additionally, the points of high porosity within these bands are recorded in both limestone and dolomite facies.

Stimulation and Induced Seismicity

A linear extrapolation of the Hickman et al. (1985) data set to the Quackenbush Hill field predicts a SH,max of 85 MPa and SH,min of 55 MPa (Table 2). Two ISIP data points are available for Quackenbush Hill field at ∼3 km depth, 52 and 58 MPa, which equate to 3 MPa of error on SH,min, or ∼5%. The calculated corresponding error on SH,max is 4 MPa.

Three wells in north-central Pennsylvania were drilled directionally at N153°E, N155°E, and N167°E (Fig. 2; American Petroleum Institute [API] numbers 117-21426, 117-20391, and 117-20330) at depths of ∼1.5 km (www.marcellusgas.org [last accessed June 2014]). Based on the assumption that operators oriented the horizontal wells perpendicular to SH,max, the inferred maximum horizontal stress at that location is oriented approximately N68°E. These data suggest that the stress field rotates counterclockwise from Auburn, New York (N83°E), toward north central Pennsylvania. A linear interpolation of stress orientation between Auburn and northern Pennsylvania indicates that SH,max at Quackenbush Hill field is approximately N71°E ± 15°. This stress vector is used for the following stimulation and seismic risk calculations that follow.

Given an assumed sliding friction coefficient (μ) of 0.85 for rocks with failure planes experiencing a normal stress of less than 200 MPa (or shallower than 5 km; Byerlee, 1978), ∼46.8 ± 5 MPa of fluid pressure would be required to hydroshear optimally oriented preexisting fractures (Fig. 8). Applying the principles of slip along preexisting planes, any fractures oriented vertically (dip of 90°) and striking approximately N045°E and N097°E ± 15° would reactivate under stimulation pressure (Fig. 9). As shown in Figure 8, the graben-bounding faults would require 5.2 MPa of additional fluid pressure in order to be reactivated, assuming there is no local reorientation of the stress field. Hydrostatic gradient (∼10 MPa/km) in a 3-km-deep well creates 30 MPa of hydrostatic pressure, leaving ∼16.8 ± 5 MPa of well-head pressure required for stimulation of preexisting fractures in the reservoir. Finally, the slip-tendency analysis (Equation 3) yields ∼2.8 MPa of shear stress and 25.3 MPa of effective normal stress (assuming 30 MPa of hydrostatic pore-fluid pressure, as in the previous analysis) on the two sets of graben-bounding faults, where θ = 84°, for a slip tendency ratio of 0.11. However, if stimulation were undertaken by adding 16.8 MPa to the in situ pore pressure, the effective normal stress would then be 8.5 MPa, which results in a slip-tendency ratio of 0.33.

Thermal Energy Resource

At the location of the Quackenbush Hill field, the top of the Black River Formation lies at an average depth of 3 km below the local surface elevation, where the average estimated temperature is 91 °C. Figure 10 illustrates the relationship between temperature and depth at the location of the dolomite in the Black River Formation among the New York T-BR fields.

Because the Trenton–Black River reservoir analysis showed both fractures and vuggy porosity, we assigned a recovery factor range that spans fractured and porous reservoir recovery, 0.05–0.25 (Williams, 2007; Williams et al., 2008). Parameters used include an approximate limestone density of 2600 kg/m3 (Manger, 1963), limestone heat capacity of 910 J/kg/°C (Robertson, 1988), water heat capacity of 4180 J/kg/°C, reinjection temperature of 50 °C, and an average reservoir temperature of 91 °C (Fig. 6). Water density was approximated at 965 kg/m3 assuming the water reaches equilibrium temperature with the rock. With those inputs, the total reservoir energy in place (Equation 5) for Quackenbush Hill is calculated as 1.3 × 1017 J. Assuming a mean recovery factor of 15% and a constant system efficiency of 90%, the estimated thermal capacity of the system is 18.4 MWth. Our conservative (recovery factor of 5%) estimate of recoverable energy from the geothermal system (Equation 7) is 6.1 MWth, and an optimistic (recovery factor of 25%) recoverable energy of 30.6 MWth.

DISCUSSION

Reservoir Quality

The Whiteman core demonstrates that vertical permeability is not sufficient for geothermal applications; however, horizontal permeability is very good, on average. The three privately held cores from Quackenbush Hill field (Marner et al., 2008) are reported to show permeability averages that are two to four orders of magnitude lower than from the Whiteman #1 core that was analyzed in this study. We believe this discrepancy may be a result of (1) inherent reservoir quality differences of the two fields; (2) sampling bias of the very limited number of cores from both fields due to the heterogeneity of the reservoirs, indicating that Quackenbush Hill may also have regions of higher permeability; or (3) both. In addition, little is known about the conditions under which the Quackenbush Hill cores were tested for permeability, or to what extent the data were corrected for the Klinkenberg effect. These factors may also have an effect on the apparent discrepancy.

Figure 6 illustrates that for the Whiteman #1 core, there is a large difference between fracture porosity and vuggy porosity, with much lower porosity for rocks containing fractures than for vuggy porosity. This suggests that the reservoir rock may be host to a dual-porosity regime, which would contribute to the apparent heterogeneity of the reservoir quality. The bimodal distribution of porosity values seen in the NPHI log (Fig. 11) also supports this hypothesis, where peaks are recorded at 0.5% and 8% porosity, and the mean porosity is 4%. The core data also suggest that isolated fracture permeability is not sufficient for geothermal applications, except where fractures coincide with vugs. Furthermore, isolated small vugs also do not have sufficient permeability (<100 mD), except where they coincide with medium and/or large vugs.

When Equation 8 was applied to the Quackenbush NPHI log data, the maximum theoretical permeability of Quackenbush Hill field is 36 D, which is three times larger than the maximum recorded permeability in the Whiteman #1 core but still in the same order of magnitude. The average calculated permeability from the log data was 120 mD, though the spread is very large, spanning 13 orders of magnitude. The spread of porosity values measured by the NPHI log (0.002%–53.4%) is also wider than the range measured in the core (0.5%–26.9%). These wide spreads are expected, given the heterogeneity of the reservoir rock and the larger sample size of the logs compared to the core. When porosity is plotted with total vertical depth (Fig. 7), three zones of high porosity stand out at discrete depths, rather than being spread throughout the entire thickness of reservoir rock. These zones could be representative of multiple vertically partitioned high-porosity zones or a single high-porosity zone that dips southward with the top of the Black River Formation. Either way, this vertical anisotropy is consistent with the previous result that horizontal permeability dominates over vertical permeability. And because there are high concentrations of low-porosity measurements within these high-porosity bands, we hypothesize that there is horizontal anisotropy as well, creating zones of high porosity that are likely concentrated in clusters, as seen in other examples of HTD reservoirs (Dewit et al., 2012).

Based on what is known of the origin and current reservoir quality of the Trenton–Black River HTD fields, we propose that fluid flow through an unstimulated Quackenbush Hill geothermal reservoir would most likely be controlled by fracture-connected vugs and clusters of variably sized vugs. However, the degree to which these clusters of high porosity and permeability are connected on a reservoir scale is currently unknown. A study by Philip et al. (2005) modeled fluid flow through a fractured dolostone reservoir and found that reservoirs with poorly interconnected fractures exhibit increased flow through the reservoir matrix (pore system). In a geothermal setting, this could equate to better heat sweep across a dolostone reservoir. The implementation of tracer tests at Quackenbush Hill field would help to characterize the connectivity between wells, reservoir geometry, and heat-sweep quality. Furthermore, geothermal fluid flow in these fields is likely to be confined to the dolomite regions, since tight limestone surrounds the dolomitized zones, which would conveniently prevent fluid loss and reduce pumping costs.

Stimulation

It is possible that the permeability in the Quackenbush Hill field, which was useful to the oil and gas industry, will not be suitable for geothermal heat extraction. In the case that the matrix permeability at Quackenbush Hill is too low to connect the clusters of high permeability found in vugs and fractures, then stimulation of the rock utilizing the principles of hydraulic shearing of preexisting fractures may be an option. Given the current stress regime, the first fractures to reactivate under increased pore-fluid pressure are oriented vertically and strike approximately N045°E and N097°E ± 15° at an estimated well-head pressure of 17 ± 5 MPa. Reactivation of such fractures would be favorable if the reservoir design goals included creating a network of fractures that strike obliquely to the strike of the grabens, forming serpentine pathways between the injection and production wells for increased heat sweep (Fig. 9). Additionally, reopening vertical fractures in the reservoir may enhance the vertical permeability, thereby optimizing heat sweep by gaining access to a greater vertical volume of rock. Stimulation may also be a feasible option for other nearby T-BR fields because their orientation (Fig. 2) relative to the known regional stress field is approximately the same as that of Quackenbush Hill.

Induced Seismicity

The principle of slip-tendency analysis states that if the ratio of shear stress to effective normal stress on a fracture is greater than or equal to the sliding friction coefficient, slip is likely to occur (Morris et al., 1996). The slip-tendency ratio for the graben-bounding normal faults under hydrostatic, non-stimulation conditions is 0.11, which is nearly seven times lower than the internal friction coefficient of 0.85 (Byerlee, 1978), indicating that the large observed faults are not close to failure in the current stress regime. Under stimulation conditions, that ratio increases to 0.33, which is still less than half of the sliding friction coefficient, indicating low risk. However, there is large uncertainty associated with the stress tensors used for this analysis, which is important to acknowledge. The confidence in the validity of the computed stress magnitudes is high because Quackenbush Hill field well ISIP values provided a reasonable upper and lower bound to our approximation. However, given the necessity of assuming a regional stress field at Quackenbush Hill field based on extrapolation from measurements at shallower depths in Auburn, New York, and northern Pennsylvania, local stress-field measurements are needed at Quackenbush Hill field. The simple slip-tendency calculation on the main graben-bounding Quackenbush faults predicts that the faults are at low risk for reactivation, with or without stimulation. However, this analysis does not take into consideration any smaller sets of subsidiary faults associated with the negative flower structures, of which the orientations are unknown. To more rigorously assess the potential for reactivation of faults at Quackenbush Hill field, it would be advisable (1) to conduct geophysical surveys that could image the orientations and lateral dimensions of large faults that are not currently described by available data; (2) to better document three-dimensional permeability near faults; and (3) to determine more accurately the local stress orientation with mini-frac tests. Public perception of induced seismicity, even of low magnitude (M1–M2) earthquakes that cannot damage surface infrastructure, is an important issue. Though we do not expect induced seismicity based on these preliminary results, seismic monitoring systems are highly advised if a geothermal energy production project was undertaken at Quackenbush.

Thermal Energy Utilization

An average reservoir temperature of 91 °C at an average depth of 3 km in the Quackenbush Hill field is adequate for a variety of end-use applications, including geothermal district heating, refrigeration, animal husbandry, greenhouses, swimming pool heating, and fish farming (Lindal, 1973). Residential use of the heat would increase infrastructure costs for piping and decrease efficiency because of the need to transport the heat to Elmira (15 km from center of field) or Corning (5 km from center of field). However, residential heating demand is high; according to the Energy Information Agency (2009), New York residences consumed 1.9 kWth on average for space heating in the year 2009. In the case that the heat-transport distance from Quackenbush Hill to either Corning or Elmira is economically feasible, the Quackenbush Hill field geothermal district heating system could provide space heating for 3200–16,100 homes over the 30-year lifetime of the reservoir.

On-site use of the hot water would require capital costs for a new greenhouse, farm, or factory but would reduce operational costs and increase the overall efficiency of the system for the lifetime of the project. The strategy of building infrastructure at the site of the well field for on-site heat utilization would also open up the possibility of harnessing heat at T-BR fields that are farther away from population centers.

CONCLUSIONS

Our feasibility study of the Quackenbush Hill field shows promise for geothermal heating applications, given its proximity to a dense population of end users. An initial assessment suggests that this field could potentially provide between 6.1 and 30.7 MWth of energy to the inhabitants and industries of the Elmira and Corning region for 30 years. Favorable conditions that are well understood and constrained by local data include temperature averaging 91 °C for direct-use applications at no more than 3.4 km depth and tight limestone sealing of a reservoir with suitable permeability, averaging 120 mD. Fluid flow through the entirety of the reservoir and between existing wells is less well known, though favorable stimulation conditions would optimize potential heat sweep, if needed. A preliminary analysis of the risk of inducing seismicity, using low-resolution regional stress data combined with minimal data on reservoir fault orientations, shows a low risk for inducing seismicity along major known faults. High-resolution stress and fault data are needed in future studies to reduce the uncertainty on this analysis.

These results may be applicable to the other New York T-BR hydrothermal dolomite gas fields. The T-BR gas fields that are still producing gas are nearing the end of production. If a relationship is built with the gas companies who currently own the fields, purchasing of the wells and additional data may be possible. Practical first steps to acquire higher-resolution data at Quackenbush Hill field are seismic surveys, pump tests, and tracer tests to better map the field architecture and fault orientations, in addition to mini-frac tests to constrain the stress state in the reservoir.

Similar feasibility analyses can be performed on other conventional hydrocarbon fields to determine their suitability for repurposing to geothermal reservoirs. For hydrocarbon fields with sufficient permeability, heat availability, and a close proximity to potential end users of geothermal hot water, there is potential for cost savings and risk reduction for the geothermal industry in the transition from depleted hydrocarbon fields to geothermal heat production. Given the similarities in operations, infrastructure, and knowledge base between the petroleum and geothermal industries, there is also an opportunity for collaboration toward a beneficial relationship with mutual goals.

The authors are grateful to Brian Slater of the New York State Museum for his help in navigating the ESOGIS website and for kindly sharing his knowledge and data from the T-BR fields. We would also like to thank Dr. Robert Jacobi for sharing his detailed posters and knowledge from previous work on the T-BR fields, Professor Katie Keranen for her expertise on induced seismicity, Professor Richard Allmendinger for his insights and for the use of his MohrPlotter and Stereonet software, and Don Fox for his reservoir engineering wisdom. We appreciate the generosity of Petra IHS for providing free software licensing for university research. A special acknowledgment is warranted for reviewer Filipe Aron, Guest Associate Editor Anna M. Crowell, and named reviewer Maria Richards, whose critiques of a previous manuscript led to significant improvements. Support for Erin Camp was provided by the National Science Foundation’s Integrative Graduate Education and Research Training program (IGERT) through an award to Cornell University (DGE-0966045).