Documents

Wiley-Contaminant pathways fr hydraulically fract shale (1).pdf

Categories
Published
of 11
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Description
Potential Contaminant Pathways from Hydraulically Fractured Shale to Aquifers by Tom Myers Abstract Hydraulic fracturing of deep shale beds to develop natural gas has caused concern regarding the potential for various forms of water pollution. Two potential pathways—advective transport through bulk media and preferential flow through fractures—could allow the transport of contaminants from the fractured shale to aquifers. There is substantial geologic evidence that natural
Transcript
  Potential Contaminant Pathways fromHydraulically Fractured Shale to Aquifers by Tom Myers Abstract Hydraulic fracturing of deep shale beds to develop natural gas has caused concern regarding the potential forvarious forms of water pollution. Two potential pathways—advective transport through bulk media and preferentialflow through fractures—could allow the transport of contaminants from the fractured shale to aquifers. Thereis substantial geologic evidence that natural vertical flow drives contaminants, mostly brine, to near the surfacefrom deep evaporite sources. Interpretative modeling shows that advective transport could require up to tens of thousands of years to move contaminants to the surface, but also that fracking the shale could reduce that transporttime to tens or hundreds of years. Conductive faults or fracture zones, as found throughout the Marcellus shaleregion, could reduce the travel time further. Injection of up to 15,000,000 L of fluid into the shale generateshigh pressure at the well, which decreases with distance from the well and with time after injection as the fluidadvects through the shale. The advection displaces native fluids, mostly brine, and fractures the bulk mediawidening existing fractures. Simulated pressure returns to pre-injection levels in about 300 d. The overall systemrequires from 3 to 6 years to reach a new equilibrium reflecting the significant changes caused by fracking theshale, which could allow advective transport to aquifers in less than 10 years. The rapid expansion of hydraulicfracturing requires that monitoring systems be employed to track the movement of contaminants and that gaswells have a reasonable offset from faults. Introduction The use of natural gas (NG) in the United States hasbeen increasing, with 53% of new electricity generatingcapacity between 2007 and 2030 projected to be with NG-fired plants (EIA 2009). Unconventional sources accountfor a significant proportion of the new NG available tothe plants. A specific unconventional source has beendeep shale-bed NG, including the Marcellus shale primar-ily in New York, Pennsylvania, Ohio, and West Virginia(Soeder 2010), which has seen over 4000 wells devel-oped between 2009 and 2010 in Pennsylvania (Figure 1).Unconventional shale-bed NG differs from conventional Hydrologic Consultant, 6320 Walnut Creek Road, Reno,NV 89523; (775) 530-1483; fax: (775) 530-1483; tom_myers@charter.netReceived August 2011, accepted February 2012. © 2012, The Author(s)GroundWater © 2012,NationalGroundWaterAssociation.doi: 10.1111/j.1745-6584.2012.00933.x sources in that the host-formation permeability is so lowthat gas does not naturally flow in timeframes suitable fordevelopment. Hydraulic fracturing (fracking, the industryterm for the operation; Kramer 2011) loosens the forma-tion to release the gas and provide pathways for it to moveto a well.Fracking injects up to 17 million liters of fluidconsisting of water and additives, including benzene atconcentrations up to 560 ppm (Jehn 2011), at pressuresup to 69,000 kPa (PADEP 2011) into low permeabilityshale to force open and connect the fractures. This isoften done using horizontal drilling through the middleof the shale with wells more than a kilometer long. Theamount of injected fluid that returns to the ground surfaceafter fracking ranges from 9% to 34% of the injected fluid(Alleman 2011; NYDEC 2009), although some would beformation water.Many agency reports and legal citations (DiGiulioet al. 2011; PADEP 2009; ODNR 2008) and peer-reviewed articles (Osborn et al. 2011; White and Mathes 872 Vol. 50, No. 6–GROUND WATER–November-December 2012 (pages 872–882) NGWA.org  Figure 1. Location of Marcellus shale in the northeastern United States. Location of Marcellus wells (dots) drilled from July2009 to June 2010 and total Marcellus shale wells in New York and West Virginia. There are 4064 wells shown in Pennsylvania,48 wells in New York, and 1421 wells in West Virginia. Faulting in the area is documented by PBTGS (2001), Isachsen andMcKendree (1977), and WVGES (2011, 2010a, 2010b). 2006) have found more gas in water wells near areasbeing developed for unconventional NG, documenting thesource can be difficult. One reason for the difficulty isthe different sources; thermogenic gas is formed by com-pression and heat at depth and bacteriogenic gas is formedby bacteria breaking down organic material (Schoell1980). The source can be distinguished based on bothC and H isotopes and the ratio of methane to higher chaingases (Osborn and McIntosh 2010; Breen et al. 2007).Thermogenic gas can reach aquifers only by leaking fromthe well bore or by seeping vertically from the source.In either case, the gas must flow through potentially verythick sequences of sedimentary rock to reach the aquifers.Many studies which have found thermogenic gas in waterwells found more gas near fracture zones (DiGiulio et al.2011; Osborn et al. 2011; Breen et al. 2007), suggestingthat fractures are pathways for gas transport.A pathway for gas would also be a pathway for flu-ids and contaminants to advect from the fractured shale tothe surface, although the transport time would be longer.Fracking fluid has been found in aquifers (DiGiulio et al.2011; EPA 1987), although the exact source and pathwayshad not been determined. With the increasing developmentof unconventional NG sources, the risk to aquifers couldbe increasing. With so little data concerning the movementof contaminants along pathways from depth, either fromwellbores or from deep formations, to aquifers, conceptualanalyses are an alternative means to consider the risks.The intent of this study is to characterize the risk factors associated with vertical contaminant transportfrom the shale to near-surface aquifers through naturalpathways. I consider first the potential pathways forcontaminant transport through bedrock and the necessaryconditions for such transport to occur. Second, I estimatecontaminant travel times through the potential pathways,with a bound on these estimates based on formationhydrologic parameters, using interpretative MODFLOW-2000 (Harbaugh et al. 2000) computations. The modelingdoes not, and cannot, account for all of the complexitiesof the geology, which could either increase or decreasethe travel times compared to those considered herein.The article also does not include improperly abandoned NGWA.org T. Myers GROUND WATER 50, no. 6: 872–882 873  boreholes which could cause rapid transport in additionto natural pathways. Method of Analysis Using the Marcellus shale region of southern NewYork (Figure 1), I consider several potential scenariosof transport from shale, 1500 m below ground surface(bgs) to the surface, beginning with pre-developmentsteady state conditions to establish a baseline andthen scenarios considering transport after fracking haspotentially caused contaminants to reach formationsabove the shale. To develop the conceptual models andMODFLOW-2000 simulations, it is necessary first toconsider the hydrogeology of the shale and the detailsof hydraulic fracturing, including details of how frackingchanges the shale hydrogeologic properties. Hydrogeology of Marcellus Shale Shale is a mudstone, a sedimentary rock consistingprimarily of clay- and silt-sized particles. It formsthrough the deposition of fine particles in a low energyenvironment, such as a lake- or seabed. The Marcellusshale formed in very deep offshore conditions duringDevonian time (Harper 1999) where only the finestparticles had remained suspended. The depth to theMarcellus shale varies to as much as 3000 m in partsof Pennsylvania, and averages about 1500 m in southernNew York (Soeder 2010). Between the shale and theground surface are layers of sedimentary rock, includingsandstone, siltstone, and shale (NYDEC 2009).Marcellus shale has very low natural intrinsic perme-ability, on the order of 10 − 16 Darcies (Kwon et al. 2004a,2004b; Neuzil 1986, 1994). Schulze-Makuch et al. (1999)described Devonian shale of the Appalachian Basin, of which the Marcellus is a major part, as containing “coalyorganic material and appear either gray or black” andbeing “composed mainly of tiny quartz grains  < 0.005 mmdiameter with sheets of thin clay flakes.” Median particlesize is 0.0069  ±  0.00141 mm with a grain size distribu-tion of   < 2% sand, 73% silt, and 25% clay. Primary poresare typically 5  ×  10 − 5 mm in diameter, matrix porosityis typically 1% to 4.5% and fracture porosity is typically7.8% to 9% (Schulze-Makuch et al. 1999 and referencestherein).Porous flow in unfractured shale is negligible dueto the low bulk media permeability, but at larger scalesfractures control and may allow significant flow. The Mar-cellus shale is fractured by faulting and contains synclinesand anticlines that cause tension cracks (Engelder et al.2009; Nickelsen 1986). It is sufficiently fractured in someplaces to support water wells just 6 to 10 km from whereit is being developed for NG at 2000 m bgs (Loyd andCarswell 1981). Conductivity scale dependency (Schulze-Makuch et al. 1999) may be described as follows: K  = Cv m K  is hydraulic conductivity (m/s),  C  is the intercept of alog-log plot of observed  K  to scale (the  K  at a samplevolume of 1 m 3 ),  v  is sample volume (m 3 ), and  m  isa scaling exponent determined with log-log regression;for Devonian shale,  C  equals 10 − 14 . 3 , representing theintercept, and  m  equals 1.08 (Schulze-Makuch et al.1999). The very low intercept value is a statistical butnot geologic outlier because it corresponds with verylow permeability values and demonstrates the importanceof fracture flow in the system (Schulze-Makuch et al.1999). Most of their 89 samples were small because thedeep shale is not easily tested at a field-scale and nogroundwater models have been calibrated for flow throughthe Marcellus shale. Considering a 1-km square area with30-m thickness, the Kh would equal 5.96  ×  10 − 7 m/s(0.0515 m/d). This effective  K  is low and the shale wouldbe an aquitard, but a leaky one. Contaminant Pathways from Shale to the Surface Thermogenic NG found in near-surface water wells(Osborn et al. 2011; Breen et al. 2007) demonstrates thepotential for vertical transport of gas from depth. Osbornet al. (2011) found systematic circumstantial evidence forhigher methane concentrations in wells within 1 km of Marcellus shale gas wells. Potential pathways includeadvective transport through sedimentary rock, fracturesand faults, and abandoned wells or open boreholes. Gasmovement through fractures depends on fracture width(Etiope and Martinelli 2002) and is a primary concern formany projects, including carbon sequestration (Annunzi-atellis et al. 2008) and NG storage (Breen et al. 2007).Open boreholes and improperly sealed water and gaswells can be highly conductive pathways among aquifers(Lacombe et al. 1995; Silliman and Higgins 1990).Pathways for gas suggest pathways for fluids andcontaminants, if there is a gradient. Vertical hydraulicgradients of a up to a few percent, or about 30 m over1500 m, exist throughout the Marcellus shale region asmay be seen in various geothermal developments inNew York (TAL 1981). Brine more than a thousandmeters above their evaporite source (Dresel and Rose2010) is evidence of upward movement from depth tothe surface. The Marcellus shale, with salinity as highas 350,000 mg/L (Soeder 2010; NYDEC 2009), maybe a primary brine source. Relatively uniform brineconcentrations over large areas (Williams et al. 1998)suggest widespread advective transport. The transitionfrom brine to freshwater suggests a long-term equilibriumbetween the upward movement of brine and downwardmovement of freshwater. Faults, which occur throughoutthe Marcellus shale region (Figure 1) (Gold 1999), couldprovide pathways (Konikow 2011; Caine et al. 1996)for more concentrated advective and dispersive transport.Brine concentrating in faults or anticline zones reflectspotential preferential pathways (Wunsch 2011; Dresel andRose 2010; Williams 2010; Williams et al. 1998).In addition to the natural gradient, buoyancy wouldprovide an additional initial upward push. At TDS equalto 350,000 mg/L, the density at 25  ◦ C is approximately 874 T. Myers GROUND WATER 50, no. 6: 872–882 NGWA.org  1290 kg/m 3 , or more than 29% higher than freshwater.The upward force would equal the difference in weightbetween the injected fluid and displaced brine. As anexample, if 10,000,000 L does not return to the surface asflowback (Jehn 2011), the difference in mass between thevolume of fracking fluid and displaced brine is approxi-mately 3,000,000 kg, which would cause an initial upwardforce. The density difference would dissipate as the saltconcentration in the fracking fluid increases due to diffu-sion across the boundary between the fluid and the brine.In just Pennsylvania, more than 180,000 wells hadbeen drilled prior to any requirement for documentingtheir location (Davies 2011), therefore the location of many wells is unknown and some have probably beenimproperly abandoned. These pathways connect aquifersthrough otherwise continuous aquitards; overpressuriza-tion of lower aquifers due to injection near the wellpathway could cause rapid transport to higher aquifers(Lacombe et al. 1995). In the short fracking period, theregion that is overpressurized remains relatively close tothe gas well (see modeling analysis below), therefore itshould be possible for the driller to locate nearby aban-doned wells that could be affected by fracking. This articledoes not consider the potential contamination althoughunlocated abandoned wells of all types must be considereda potential and possibly faster source for contaminationdue to fracking. Effect of Hydraulic Fracturing on Shale Fracking increases the permeability of the targetedshale to make extraction of NG economically efficient(Engelder et al. 2009; Arthur et al. 2008). Frackingcreates fracture pathways with up to 9.2 million squaremeters of surface area in the shale accessible to ahorizontal well (King 2010; King et al. 2008) andconnects natural fractures (Engelder et al. 2009; Kinget al. 2008). No post-fracking studies that documentedhydrologic properties were found while researching thisarticle (there is a lack of information about pre- and post-fracking properties; Schweitzer and Bilgesu 2009), butit is reasonable to assume the  K  increases significantlybecause of the newly created and widened fractures.Fully developed shale typically has wells spaced atabout 300-m intervals (Edwards and Weisset 2011; Soeder2010). Up to eight wells may be drilled from a singlewell pad (NYDEC 2009; Arthur et al. 2008), althoughnot in a perfect spoke pattern. Reducing by half theeffective spacing did not enhance overall productivity(Edwards and Weisset 2011) which indicates that 300-mspacing creates sufficient overlap among fractured zonesto assure adequate gas drainage. The properties controllinggroundwater flow would therefore be affected over a largearea, not just at a single horizontal well or set of wellsemanating from a single well pad.Fracking is not intended to affect surrounding forma-tions, but shale properties vary over short ranges (King2010; Boyer et al. 2006) and out-of-formation fracking isnot uncommon. In the Marcellus shale, out-of-formationfracks have been documented 500 m above the top of theshale (Fisher and Warpinski 2011). These fractures couldcontact higher conductivity sandstone, natural fractures, orunplugged abandoned wells above the target shale. Also,fluids could reach surrounding formations just because of the volume injected into the shale, which must displacenatural fluid, such as the existing brine in the shale. Analysis of Potential Transport along Pathways Fracking could cause contaminants to reach overlyingformations either by fracking out of formation, connectingfractures in the shale to overlying bedrock, or bysimple displacement of fluids from the shale into theoverburden. Advective transport, considered as simpleparticle velocity, will manifest if there is a significantvertical component to the regional hydraulic gradient.Numerical modeling, completed with the MODFLOW-2000 code (Harbaugh et al. 2000), provides flex-ibility to consider potential conceptual flow scenarios, butshould be considered interpretative (Hill and Tiedeman2007). The simulation considers the rate of vertical trans-port of contaminants to near the surface for the differentconceptual models, based on an expected, simplified, real-istic range of hydrogeologic aquifer parameters.MODFLOW-2000 is a versatile numerical modelingcode, but there is insufficient data regarding the geologyand water chemistry between aquifers and the deep shale,such as salinity profiles or data concerning mixing of thebrine with fracking fluid, to best use its capabilities. Asmore data becomes available, it may be useful to considersimulating the added upward force caused by the brine byusing the SEAWAT-2000 module (Langevin et al. 2003).Vertical flow would be perpendicular to the generaltendency for sedimentary layers to have higher horizontalthan vertical conductivity. Fractures and improperlyabandoned wells would provide pathways for muchquicker vertical transport than general advective transport.This article considers the fractures as vertical columnswith model cells having much higher conductivity thanthe surrounding bedrock. The cell discretization is fine, sothe simulated width of the fracture zones is realistic. Dualporosity modeling (Shoemaker et al. 2008) is not justifiedbecause turbulent vertical flow through the fractures isunlikely, except possibly during the actual fracking thatcauses out-of-formation fractures, a scenario not simulatedhere. MODFLOW-2000 has a module, MNW (Halfordand Hanson 2002), that could simulate rapid transportthrough open bore holes. MNW should be used insituations where open boreholes or improperly abandonedwells are known or postulated to exist.The thickness of the formations and fault would affectthe simulation, but much less than the several-order-of-magnitude variation possible in the shale properties. Theoverburden and shale thickness were set equal to 1500 and30 m, respectively, similar to that observed in southernNew York. The estimated travel times are proportionalfor thicker or thinner sections. The overburden couldbe predominantly sandstone, with sections of shale,mudstone, and limestone. The vertical fault is assumed NGWA.org T. Myers GROUND WATER 50, no. 6: 872–882 875
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks