Will water scarcity in semiarid regions limit hydraulic fracturing of shale plays?

Unconventional O&G production in the Eagle Ford Play (∼9500 mi², 24 600 km² area) began in 2008 (figure 1). Water use for HF in the play totaled 40 bgal (150 bL) for 8301 wells (2009–2013) [6].

Zoom In Zoom Out Reset image size

Precipitation in the Eagle Ford play ranges from 19.5 inches/yr (495 mm yr⁻¹) in the west to 38.6 inches/yr (980 mm yr⁻¹) in the east (figure S1). Total population in the play area was ∼0.2 million in 2010 in the mostly rural 16 county area. Nearby major cities include San Antonio (population 2.2 million) and Laredo (population 236 000) (figure S2). Land use/land cover consists of shrubland (45%), pasture (25%), grassland (8%), forest (7%), and cropland (5%) (figure S2). Most cropland and some pasture are irrigated. Major rivers include, from west to east, the Rio Grande, Nueces, Frio, Atascosa, San Antonio, and Guadalupe rivers. Major aquifers in the region include the Carrizo–Wilcox aquifer and the Gulf Coast aquifer system, and minor aquifers include the Queen City, Sparta and Yegua-Jackson aquifers (figures 2, S3) [23]. Groundwater Conservation Districts (8 GCDs) manage regional groundwater according to Texas Senate Bill 1, an omnibus water bill passed in 1997 (figure S4).

Zoom In Zoom Out Reset image size

HF water demand was compared with water supplies at the play and county and square mile grid levels while also considering water use in other sectors to assess vulnerability to water constraints (figure 3). Additional details on data sources are available in SI, section 6. HF water demand or use was based on water use to date (2009–2013) in the Eagle Ford [6]. Information on wells used to supply water for HF was obtained from the Texas Dept. of Licensing and Regulation. HF water use was compared with that for other sectors based on the most recent data from the Texas Water Development Board [24]. Future HF water use was estimated by assuming that the highest established well density (HEWD) for HF wells in each of the seven production zones (2 oil, 2 volatile oil condensate, 2 condensate, and 1 dry gas) or county/production zone intersections (40) would apply to the entire play (SI, section 6, tables S1, S2).

Zoom In Zoom Out Reset image size

Primary sources of HF water include surface water and groundwater from aquifer recharge, aquifer storage, and cross-formational flow from adjacent aquifers; however, the latter was neglected in this analysis. Data on water diversions from the Rio Grande for mining were obtained from the Rio Grande Water Master. Renewable groundwater supplies for HF were estimated from aquifer recharge rates using a global model (PCR GLOBWB) as used in recent studies [1, 5, 25], regional recharge rates from groundwater chloride data [26], and groundwater availability models (GAMs) [27–29]. Groundwater supplies from aquifer storage were estimated from GAMs [22, 27, 28] and include drainable storage (water that would be drained under gravity, aquifer thickness × specific yield) in aquifer outcrop zones and compressible storage (water released from compressible storage: storativity × head decline) in downdip confined sections of aquifers (figure 2, SI, section 9, figure S11).

Impacts of HF on groundwater resources were evaluated using groundwater level monitoring data from 131 wells from the Texas Water Development Board (TWDB) database. Flow duration curves and baseflow were estimated for 20 streamflow gages using the length of record to determine which streams are perennial and could be impacted by HF groundwater pumpage. Net impact of HF on water resources was examined by comparing water use for shale gas extraction with water use for cooling in natural gas power plants [30].

Alternatives to freshwater sources for HF were evaluated, including flowback/produced (FP) water (SI, section 11), brackish groundwater, and municipal waste water. FP water volumes were estimated for individual wells from 2009–2013. The ratio of FP to HF water was calculated to evaluate what percentage of HF water would be available for reuse or recycling. Brackish groundwater was estimated by integrating groundwater quality (TDS) data from TWDB with groundwater storage data from the GAMs. Availability of municipal waste water was evaluated by contacting the 20 largest municipalities in the play area. We contacted O&G operators to determine trends in HF water sourcing through the South Texas Energy and Economic Roundtable (STEER.com) and through conferences. Analyses of water demand versus supply in the Eagle Ford play were compared with water demand from literature-based estimates for the semiarid Permian Basin play in west Texas [5] and from previous analysis for the semiarid Bakken play in N Dakota and Montana [6]. Water demand in the Permian Basin was compared with water supplies from groundwater storage, including brackish water from the GAMs and discussions with operators through the Permian Basin Petroleum Association. Water demand in the Bakken was compared with reservoir storage in the Missouri River [31].

Water demand or use for HF in the Eagle Ford in 2013 (17.8 bgal, 67 bL) is assumed to be mostly consumptive because of limited recycling as discussed later. At the state level, HF water demand in the Eagle Ford in 2013 represents 0.3% of statewide water withdrawal (5200 bgal, 20 000 bL, 2012) and 0.5% of statewide water consumption (3900 bgal, 15 000 bL, 2012), assuming 85% of irrigation and 40% of municipal water use is consumptive [32]. The latest year with reported state water use data is 2012. At the play level, HF water consumption represents 4–9% of water withdrawal in 2011 and 2012 and 5–11% of water consumption (figure 5, table S2). However, using 2013 HF water use with 2012 water use data results in HF representing 13% of water withdrawal and 16% of water consumption. The dominant water user in the play is irrigation, accounting for 62–65% of water consumption (2011–2012), followed by municipal (10–12%), and steam electric power (7–13%) (figure 5).

Zoom In Zoom Out Reset image size

Total HF drilling during the remaining life of the play (∼20 years) was estimated based on the current highest established well density (HEWD) to be 50 700–70 400 wells, resulting in ∼42 000–62 000 wells remaining after subtracting the 8300 wells drilled to date [6] (table S3; figure S6), bounding the previous estimate of 50 000 wells [8]. Multiplying 2013 drilling (3512 wells) by 20 and subtracting 8300 would result in a similar estimate (62 000 wells). These projections would result in ∼223–333 bgal (844–1260 bL) of additional HF water based on average 2013 HF water use/well in each production zone (oil, volatile oil, condensate, and dry gas) (table S3), representing ∼6–8 times HF water use to date (∼40 bgal, 150 bL, 2009–2013) [6]. To assess water constraints for HF, we used the upper bound for HF water demand (∼330 bgal, 1250 bL) as a conservative estimate (table 1).

The dominant HF water supplies in the Eagle Ford play include aquifer recharge and freshwater storage (table 1). The focus on freshwater derives in part from reports that many landowners stipulate use of their fresh groundwater as part of lease agreements to generate an additional revenue stream to land owners and the sole revenue stream to landowners with severed mineral rights. Long-term (1960–2000) mean recharge rates from simulated drainage below the root zone in the PCR-GLOBWB model totaled 49 bgal yr⁻¹ (185 bL yr⁻¹) for the Eagle Ford play area (table S5), ∼170% greater than HF water demand in 2013 (18 bgal, 68 bL) but 65–72% less than total water withdrawal in the play area in 2011 (175 bgal, 662 bL) and 2012 (142 bgal, 538 bL) (table S3). Regional recharge rates from groundwater chloride data in the Carrizo–Wilcox aquifer totaled 43 bgal yr⁻¹ (163 bL yr⁻¹) [26]. Sources of recharge include diffuse recharge from percolation below the root zone and focused recharge beneath losing streams (e.g. Leona, Nueces, and Frio rivers) [33, 34]. Simulated recharge in regional GAMs in outcrop areas in the play totaled ∼60 bgal yr⁻¹ (227 bL yr⁻¹; table S5). All of these recharge estimates refer to aquifer outcrop zones; however, most O&G water supply wells in the western part of the play are in the deeper confined parts of the aquifers (1000–7000 ft deep, 300–2000 m) (figure S5). Net flow to the downdip confined aquifer was estimated to be ∼34% of the outcrop recharge for the Carrizo–Wilcox aquifer, the dominant aquifer in the west, based on the GAM [26]. Applying the percentage of deep recharge in the Carrizo aquifer (34%) to other confined aquifers in the play as an estimate results in net recharge of ∼20 bgal yr⁻¹ (76 bL yr⁻¹), which is similar to HF water use in 2013 (17.8 bgal) but is ∼86–89% less than total water withdrawal in 2011 and 2012 in the play area. Therefore, relying on recharge alone for HF is inadequate because recharge is not uniformly distributed in the play (figure S10) and most of the HF water supply wells are in the deep confined aquifers with very low recharge rates.

The largest water source for HF is groundwater storage in the play area. A previous study of water sources for irrigation pumpage in the confined portion of the Carrizo–Wilcox aquifer indicates that ∼60% of the pumped water was derived from aquifer storage, which is not sustainable and has resulted in groundwater depletion [21]. Freshwater extends far downdip in the Carrizo aquifer in the western part of the play, reflecting high permeability sands but is much more limited in the east [35] (figures 2(b), (c), S9). Drainable freshwater storage in the unconfined portions of the aquifers, estimated from integration of TDS data (figures S19–S23) with storage from the GAMs, totaled ∼6000 bgal (22 700 bL, tables 1 and S6); however, most HF water supply wells are completed in the confined portions of the aquifers, with compressible freshwater storage of ∼3800 bgal (14 400 bL) in the play area (table S6). Projected total HF water demand (330 bgal, 1200 bL) represents ∼3% of freshwater storage (9800 bgal, 37 000 bL) at the play level (table S4). The projected 20 yr water demand from all sectors (∼1400 bgal based on average 2000–2012 withdrawals) is mostly irrigation (62%) and municipal (12%) use, and represents ∼14% of total freshwater storage in the play (table S4). Projected HF water demand at the county level ranges from <1–27% of total freshwater storage (figure 6, table S4). Projected water withdrawal for all sectors at the county level represents 1–36% of total freshwater storage for all counties except Frio (111%) and Zavala (262%) counties which are dominated by irrigation (90–95% of county water use).

Zoom In Zoom Out Reset image size

A secondary source of water for HF is the Rio Grande for counties adjacent to the river, mostly Webb County. Diversions from the Rio Grande for mining in 2011–2013 ranged from 1.1 to 2.3 bgal yr⁻¹ (4.2–8.8 bL yr⁻¹; table S8). Assuming this water is available for HF, it could provide HF water for ∼230–500 wells/yr assuming an average water use of 4.8 × 10⁶ gal/well, 18.2 × 10⁶ L/well [6] (figure S12). However, there is limited drilling near the Rio Grande and operators do not normally transport water more than ∼5–10 miles.

The 2011 drought was the most extreme on record for Texas, with precipitation in the play only 42% of the long-term mean (12.3/29.3 inches, www.prism.oregonstate.edu). Droughts generally have a much greater impact on surface water than groundwater; however, reported diversions from the Rio Grande for mining for 2011 were similar to those in 2012 and 2013 (table S8). Drought should not markedly impact water demand for HF or groundwater supply, particularly for the confined aquifers which have been recharging for several thousand years [36]. The main impact of the 2011 drought was related to increased water demand for irrigation (figure 5, table S2). The increase in water withdrawal related to irrigation from 2010 to 2011 (33 bgal, 125 bL) due to the 2011 extreme drought is similar to total HF water use for 2012 and 2013 combined (30 bgal, 114 bL). However, the main irrigation region, the Winter Garden region, is mostly north of the Eagle Ford play area and generally does not compete with HF water use (figures 1 and 4).

In addition to physical availability of water, legal access to water and regulations governing water resources need to be considered. Legal challenges relate to groundwater ownership in Texas, with current law providing that landowners have a 'vested property right' to groundwater in place [37]. Therefore, regulations, including reduced permits by Groundwater Conservation Districts that would restrict landowners' access to groundwater, may constitute 'a taking' and would have to be compensated [38]. These recent rulings call into question the scope and authority of GCDs. In addition, because of confusing language in the Texas Water Code (SI, section 9), it is not clear whether groundwater produced for HF activities is exempt from GCD regulation. While groundwater use for O&G exploration in Texas is exempt from regulation, confusion results from whether to consider HF part of exploration or production.

Groundwater management by the GCDs is based on estimates of desired future conditions (DFCs) for aquifers within groundwater management areas that result from joint-planning efforts by multiple GCDs (Texas Water Code §36.108). In Groundwater Management Area 13, covering most of the Eagle Ford play (figure S4), the desired future condition is a not-to-exceed area-average drawdown of 23 ft (7 m) across the GCD by 2060. However, localized drawdown in pumping centers will likely exceed this amount. This desired future condition translates to modeled available groundwater of ∼150 bgal yr⁻¹ (570 bL yr⁻¹) from the GAMs to honor the drawdown limitation and planning horizon. The 2013 HF water use (18 bgal, 68 bL) corresponds to ∼12% whereas total withdrawal from all sectors (142 bgal, 538 bL) represents ∼95% of the annual modeled available groundwater in this region. These desired future conditions will be revised in 2016 and should reflect groundwater pumpage for HF in the region.

Groundwater level monitoring data from 131 wells in the play area indicate that water levels have declined to a maximum depth of ∼200 ft (∼60 m) in the west and mostly ≤50 ft (15 m) in the east (figures 7, 8, and S13). Greater declines in the west may be related to the deeper O&G water supply wells completed mostly in confined aquifers relative to shallower wells in the east, many of which may be completed in unconfined aquifers (figure S5). The specific yield for most of the unconfined aquifers is ∼0.15 whereas the average storativity of the confined Carrizo–Wilcox aquifer in the west is 0.0012 (table S7). The ratio of the volumetric extent of cones of depression in confined versus unconfined aquifers is equivalent to the ratio of specific yield to storativity, about a factor of 1000 in many cases [39] (SI, section 9). A crude estimate of the impacts of HF water use to date (2009–2013) on groundwater levels in the confined Carrizo–Wilcox aquifer in the west was based on the assumption that all HF water was sourced from compressible aquifer storage within a mile of HF wells (figure 7). While water-level declines are small in much of the region, with 69% of western play area having no declines, 74% ≤ 25 ft (7.6 m) decline, and 81% ≤ 50 ft (15 m) decline, large declines (mostly 100–200 ft, 30–60 m) were found in 6% of the land area (figure 7). These estimates of water-level declines are generally consistent with the limited water-level monitoring data and coincident with high density of HF wells.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Impacts of past irrigation pumpage can be used as an extreme example of potential impacts of future HF water use in the region. Irrigation in the Winter Garden district lowered groundwater levels by up to 330 ft (100 m) regionally during the past century, decreasing water storage by ∼4000 bgal (15 000 bL) in the Carrizo–Wilcox aquifer (figure S14). Irrigation pumpage reversed groundwater discharge in much of the region and changed some rivers (Atascosa, Frio, and San Antonio rivers) from gaining to losing [21]. No land subsidence has been recorded in this region in response to groundwater depletion, most likely because the sediments are generally indurated. Projected HF water use over the 20 yr life of the play is ∼10% of groundwater depletion from past irrigation. The lack of subsidence from past depletion suggests that HF pumpage should not cause subsidence. HF pumpage may not impact rivers in the west, which are already losing, but could impact river flow in the east which are mostly perennial (figure S15, table S9).

Previous studies emphasize the net impact of HF water use on water resources using a LCA of gas production [16, 40, 41]. The primary market for natural gas is thermoelectric power generation, with natural gas representing ∼50% of power generation in Texas in 2012 (SI, section 10, figure S16). Most natural gas power plants in Texas have combined cycle generators (NGCC, 170 out of 215 TWh, 80%) with consumptive cooling water requirements totaling ∼32 bgal (121 bL) in 2012 (∼0.19 gal kWh⁻¹) [30]. Power generation at NGCC plants in Texas in 2012 (170 × 10¹² Wh) corresponds to 580 × 10¹² Btu of energy (3.412 Btu Wh⁻¹ of electricity) requiring 1318 × 10¹² Btu of natural gas assuming 44% fuel efficiency of NGCC power plants in Texas (i.e. 44% of the energy goes to electricity with the remaining 56% to waste heat). The amount of water required to extract Eagle Ford shale gas is estimated to be 1.5 gal/10⁶ Btu [6]. If the natural gas required to power all Texas NGCC power plants during 2012 had been produced entirely from the Eagle Ford, this would represent ∼2 bgal (7.6 bL) of water to extract the gas (1318 × 10¹² Btu × 1.5 gal/10⁶ Btu = ∼2.0 bgal of water). Therefore, water consumption for extracting the gas represents ∼6% of the cooling water consumed at the NGCC power plants (2.0 bgal/32 bgal), similar to the estimate of 6% for the Marcellus shale gas in Pennsylvania [40]. In addition, recent replacements of retiring coal steam turbine plants with NGCC plants, which have 1/3rd of the cooling water requirements, result in a net savings of water in the state [32, 42]. However, this water savings does not occur where the gas is produced in the west but where the power in generated, mostly in the central and eastern parts of the state.

Many suggest that the net impact of HF water use on water resources and the hydrologic cycle is much greater than that of other sectors, such as irrigation, because FP water in Texas is mostly disposed of in deep injection wells and removed from the hydrologic cycle (figure S5). However, in the Eagle Ford play, most of the water used for HF is not part of the active hydrologic cycle to begin with as it is mostly derived from deep confined aquifers that have been recharging for several thousand years [36]. In addition, previous studies indicate that methane combustion from natural gas in the Marcellus Shale play puts water vapor back into the hydrologic cycle (5.7 × 10⁶ gal/well, 22 × 10⁶ L/well) that more than compensates for HF water use (3.6 mgal/well, 14 × 10⁶ L/well) [40]. Similar calculations for the Eagle Ford indicate that water vapor from methane combustion alone based on methane production to date is equivalent to ∼90% of HF water use in the play (SI, section 10), ignoring water vapor generated by combustion of other hydrocarbons produced in the play.

Operators are considering various approaches to reduce vulnerability of HF to water constraints and reduce impacts of HF on water resources. The basic approaches include reducing water demand, increasing water supplies (particularly non-freshwater supplies), and intersectoral transfers (primarily from irrigation to HF). Water demand can be reduced by changing HF fluid types as shown by a recent analysis indicating ∼55–110% higher water demands associated with slickwater HF relative to cross-link gel HF; however, these changes in HF fluid types generally occurred in the early stages of Eagle Ford play development [6]. Water demand can also be reduced through reuse-recycling of FP water. Non-freshwater sources include brackish water and municipal waste water.

3.8.1. Flowback/produced water.

Small FP water volumes generally do not support reuse/recycling requirements. FP/HF ratios are generally <5% within the first month of well completion in all production zones, making it difficult to collect sufficient water to support recycling for HF (figures 9, S18). Most FP water is disposed of in the ∼700 active UIC Class II injection wells in the 18 county area, including and extending beyond the 16 counties in the Eagle Ford play (figure S5). The quality of the FP water is not highly saline, with reported values of ∼40 000 mg L⁻¹ TDS. Some operators are doing a limited amount of recycling; however, there is no formal reporting of these volumes. Portable treatment units are used to treat the FP water, particularly for reducing boron levels, which must be managed for the cross-link gel HF fluid additives. The logistics of recycling is complicated because of the low FP volumes. The economics of recycling is also generally not favorable with relatively low injection well disposal costs.

Zoom In Zoom Out Reset image size

To promote reuse/recycling of FP water the Texas Railroad Commission modified their rules in March 2013. The new rules allow operators to recycle FP fluids on their own leases or they can transfer those fluids for recycling to another operator without the need for a permit [43]. Some operators report that these new rules promote increased reuse/recycling of water.

3.8.2. Brackish groundwater resources

There is an estimated ∼35 000 bgal (130 000 bL) of brackish water in unconfined sections of aquifers and ∼45 000 bgal (170 000 bL) of compressible brackish groundwater storage in the confined sections of the aquifers in the Eagle Ford play area, totaling 80 000 bgal (300 000 bL, figure 10, tables 1, S4, S10). Projected 20 yr HF water demand represents 0.4% of brackish groundwater storage, suggesting sufficient water at a play level. At the county level, projected HF water use is ≤16% of total brackish water storage (table S4). Projected total water demand by all sectors represents ∼2% of brackish water storage at the play level and ranges from 0–13% of brackish water storage, with the exception of Frio (35%) and Zavala (78%) counties where irrigation is dominant and DeWitt county (62%) where brackish water is limited.

Zoom In Zoom Out Reset image size

A previous survey of operators in 2010 indicated that ∼20% of the HF water use throughout the play was brackish [44]; however, HF water use has expanded greatly since that time. To avoid impacting shallower domestic wells, many operators report drilling into deeper units to limit the potential for competition for shallow groundwater. Some operators indicate that 60–80% of their water use is brackish (e.g. Marathon Oil Company reported 85% of their water use in 2012 was non-freshwater, [45]); however, there is no formal reporting of this water source. Another operator reported flowing artesian production from a 6000 ft (1800 m) deep well in the eastern part of the play (Dewitt County) with TDS of ∼36 000 mg L⁻¹. Water treatment is often required to make the water compatible with HF fluid additives. To incentivize use of brackish groundwater, the Pecan Valley GCD recently revised their regulations to allow up to 20 times more groundwater pumpage if brackish water is withdrawn from the aquifer (SI, section 9).

3.8.3. Municipal waste water

3.8.4. Purchasing water from irrigation

The previous discussions of water demand versus supply focused on fresh and brackish water separately. However, projected HF water demand is a small percentage of total water resources at both the play (0.4%) and the county (0.1–6%) levels. Projected total water demand for all sectors (HF, irrigation, municipal and other) is also a small percentage of total available water at the play level (1.6%) as well as at the county level for most counties (0.2–7%), excluding two counties with high irrigation demands (Frio: 27%, Zavala: 60%), and one county (Maverick) that has essentially no modeled available groundwater resources in the play (table S4). Mapping the relationship between demand and supply indicates that there should not be problems with available water throughout much of the play, with the highest demands relative to supplies focused where irrigation demand is currently high in the northwest (figure 11).

Zoom In Zoom Out Reset image size

Uncertainties in water demand to date (2009–2013) are considered relatively low because of the reporting requirements and availability of multiple databases [6]. The analysis of water demand in the Eagle Ford play is hindered by lack of reporting requirements for the source of water used for HF (e.g. surface water, groundwater, recycled FP water) or the quality of the water used (fresh versus brackish). Uncertainties in projected HF water demands are considered high because they are based on the assumption that the HEWD to date will apply to entire O&G production zones. A much more granular analysis based on square mile productivity and considering production economics, such as that available for the Barnett play [7], is required to improve estimates of projected HF water demand. This level of analysis is currently being conducted for a project funded by the Sloan Foundation [47]. Uncertainties in water demand for other sectors also need to be considered and are considered high for irrigation. Current studies funded by TWDB are evaluating remote sensing approaches to improve these estimates [48].

Data on water supplies in the Eagle Ford rely heavily on GAMs for the region. These models were designed to assess mostly freshwater resources in aquifers; however, the Bigford and Laredo Fms. which are stratigraphically equivalent to the Queen City and Sparta, respectively, west of the Frio River are not considered aquifers and recharge and storage estimates are considered less reliable west of the Frio River. The emphasis of these models is on fresh water and estimates of brackish water are not as reliable because of limited data availability. Future models should be developed that focus on brackish water resources and integrate data being collected by industry in these regions. A recent analysis of geophysical logs in the Eagle Ford play supports the estimated volumes of brackish water in these regions [49] (figure S24). While comparisons of water demand versus supply in this study have been limited, future modeling would provide more comprehensive comparisons that would include all the water sources, including groundwater from recharge and fresh and brackish water storage, and leakage from adjacent geologic units. More detailed models may be required for assessing water demand versus supply at the level of municipalities. The impact of pressure loss in confined aquifers on well yields is not known in detail. In particular, how many water wells would be required to produce similar water volumes and would water production be economical?

While the current water level monitoring network has been invaluable in documenting impacts of water use in the play on groundwater depletion, this network should be expanded and should be designed to provide detailed information in critical areas, such as those in the vicinity of municipal well fields.