SPE 125740

Marcellus Shale Post-Frac Flowback Waters – Where is All the Salt Coming From and What are the Implications? M.E. Blauch, Superior Well Services,Inc.; R.R. Myers, T. R. Moore; B.A. Lipinski, Exco - North Coast Energy, Inc.; N.A. Houston, Superior Well Services, Inc.

Copyright 2009, Society of Petroleum Engineers This paper was prepared for presentation at the 2009 SPE Eastern Regional Meeting held in Charleston, West Virginia, USA, 23–25 September 2009. This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract One of the most prominent unexplained phenomena observed in the Marcellus and some other shale plays is the concentration of dissolved salts in produced waters after hydraulic stimulation. In this paper, we present both geochemical and lithologic laboratory and field data to address the salt question. Is salt being dissolved from the shale, or are deep saline aquifers being breached during hydraulic fracturing? What evidence do we have to support or refute either theory?

To address these questions, over 100 flowback analyses were collected over 18 months from both the southwestern and northeastern regions of the Marcellus Shale play. These data incorporate both cation and anion water analyses in either a full or partial determination of the cation and anion balance. Detailed inorganic geochemical and mineralogical analyses of shale samples were integrated to help determine the presence or absence of physical evidence of minerals that may be the root cause for high salinity. If present in the shale, is halite dissolution desirable from a formation stimulation perspective? If so, is this gain a worthwhile trade when balanced against the costs of returned load water handling and reuse or disposal? This paper provides interpretations at both the regional and local scales to try to explain basinal variations observed in the data. It discusses implications of the phenomenon of high saline frac flowback fluids, along with methods being used to mitigate environmental problems associated with the post-frac flowback water geochemistry.

Problem Statement Our own and citizen concerns about the potential environmental impacts on surface water purity, municipalities’ worries about possible diminishing fresh water supplies, and increased scrutiny by various regulatory agencies are drivers behind efforts by the natural gas industry to reduce both the consumptive use of fresh water by large fracture stimulation jobs and subsequent discharge of post-frac flowback fluids to the environment. This is especially the case in the Marcellus Shale play in Pennsylvania, where rapid development is occurring and deep-well disposal capacity is very scarce. Highly saline produced waters from Marcellus wells are presenting an enormous challenge to both regulators and operators.

Flowback waters from the Marcellus characteristically carry high levels of total dissolved solids (TDS) in the form of soluble chloride salts. The make-up water used to fracture the well is normally fresh. Operating and well-service companies have approached the problem of renewable freshwater supply by separating, filtering, and even distilling produced formation waters and frac-fluid flowback waters for future use or surface discharge (Weatherford Telegram 2007). What to do with the post-frac flowback waters in the light of scarce brine disposal facilities and substantial handling costs is an enormous burden to the economic development of the Marcellus natural gas resource.

In the Commonwealth of Pennsylvania, new regulatory limits have been proposed further limiting discharges. The Pennsylvania Department of Environmental Protection announced on April 15, 2009 that all industrial discharges will be limited to 500 mg/L TDS on January 1, 2011. There are currently no facilities in the state that can treat flowback fluids to this level. The options for an economic solution are few for operators in dealing with these saline flowback fluids. Evaporation/crystallization (EC), the only established technology for treatment of the produced waters that can achieve the newly proposed TDS limit, produces a very highly concentrated brine solution or large volumes of crystalline salt cake that still must be disposed. A 1 million gal/day crystallization plant will generate approximately 400 tons/day of salt waste. Unless some beneficial use for these residues can be found, they will require disposal in a secure solid waste facility. A typical municipal landfill cannot accept large volumes of crystalline salts and suitable facilities can do so only at a premium. Further, an EC plant is very energy intensive and thus has the potential for increased air quality impact and greenhouse gas emissions in addition to its cost of operation. The Marcellus shale gas industry may be left with no economically viable disposal options.

2 SPE 125740

The origin of the salt in the flowback waters has not been addressed sufficiently. If the source were known, there may be ways found to mitigate or avoid the levels of salt in flowback waters requiring disposal. Potential diagnostic information may be found in well-to-well and basin-wide analysis of geochemical data derived from post-fracturing flowback waters. Characterization of Post-Frac Flowbacks Since the impetus for this paper involves the observation of high salinity post-frac flowback water, a number of example flowback analyses are presented. These include both individual well examples and bulk receiving waters that have been sent to two different publicly operated treatment works (POTW’s) in the Commonwealth of Pennsylvania. Individual Well Examples Marcellus Shale Post-Frac Flowback Well A – Southwestern, PA Nineteen samples of post-frac flowback water were collected from a Marcellus well in southwestern Pennsylvania, identified here as Well A. The flowback volume observed was 15,000 bbl, representing a load recovery of ~40% (Fig. 1). A water analysis was run on each sample, the results compiled, and the trends from those observations presented here.

Conclusions from Well A flowback analysis are: 1. The amount of dissolved constituents increased as flowback progressed (Fig. 2). Both sodium and calcium show

similar trends in Well A. 2. Sodium and calcium are the most prevalent cations (Fig. 2). 3. Alkalinity and pH dropped as flowback progressed, potentially explaining the rise of calcium levels. 4. Sulfate scaling is likely as calcium is rising while sulfate is dropping (Figs. 2 and 3). 5. Fig. 4 shows a monovalent ion trend; Fig. 5 illustrates a divalent cation trend. 6. As shown in Fig. 6, the sharp rise in barium levels in the latter stages of flowback (at about 30% load recovery)

suggests potential barium sulfate scale formation during the last portion of the flowback. The solubility of barium sulfate is very low and it can be a very aggressive scale.

7. Iron content in the flowback increased as flowback progressed (Table 1). 8. Chemical composition of these waters can be classified as highly saline. 9. With cations usch as Mg, Sr, and Ba, chemical signatures of the waters are consistent with an evaporite- and

carbonate-rich source.

0.0%

10.0%

20.0%

30.0%

40.0%

0 5,000 10,000 15,000

Load

 Recovery (%

 total)

Flowback (barrels) Fig. 1—Marcellus Shale Well A flowback–load volume percent recovery vs. flowback volume.

SPE 125740 3

0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

40,000

0 5,000 10,000 15,000

Constitue

nt (p

pm)

Flowback (barrels)

Sodium

Calcium

Fig. 2—Marcellus Shale Well A flowback analysis–major cation trend. 

0

200

400

600

800

1,000

1,200

0

10,000

20,000

30,000

40,000

50,000

60,000

70,000

80,000

0 5,000 10,000 15,000

Sulfa

tes (pp

m)

Chloride

s (pp

m)

Flowback (barrels)

Chlorides Sulfates

Fig. 3—Marcellus Shale Well A flowback analysis–anion trend.

4 SPE 125740

0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

40,000

0 5,000 10,000 15,000

Constitue

nt (p

pm)

Flowback (barrels)

Sodium

Potassium

Fig. 4—Marcellus Shale Well A flowback–monovalent ion trend. 

0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

9,000

10,000

0 5,000 10,000 15,000

Constituen

t (pp

m)

Flowback (barrels)

Calcium

Magnesium

Strontium

Fig. 5—Marcellus Shale Well A flowback–divalent cation trend. 

SPE 125740 5

0

10

20

30

40

50

60

70

80

90

100

0 5,000 10,000 15,000

Constituen

t (pp

m)

Flowback (barrels)

Barium

Fig. 6— Marcellus Shale Well A flowback–barium trend. 

TABLE 1—MARCELLUS SHALE WELL A LATE STAGE FLOWBACK WATER CHEMICAL CHARACTERIZATION DATA Flowback (bbl): 12,000 bbl 13,000 bbl 14,000 bbl 15,000 bbl

Anions P Alkalinity (mg/L as CaCO3) 0 0 0 0 M Alkalinity (mg/L as CaCO3) 280 240 200 160 Chloride (mg/L as Cl-) 54,000 59,000 62,900 67,800 Sulfate (mg/L as SO4

2-) 31 20 20 24 Cations Sodium (mg/L as Na1+) 26,220 28,630 31,810 35,350 Potassium (mg/L as K1+) 1,119 1,201 1,350 1,480 Calcium (mg/L as Ca2+) 7,160 7,680 8,880 9,720 Magnesium (mg/L as Mg2+) 341 463 488 805 Total Hardness (mg/L as CaCO3) 19,300 21,100 24,200 27,600 Barium (mg/L as Ba2+) 28.9 43.3 99.6 175.7 Strontium(mg/L as Sr2+) 1,110 1,305 1,513 1,837 Iron, Ferrous (mg/L as Fe) 0.4 0.9 1.1 3.3 Iron, Total (mg/L as Fe) 63 66 72 78 Miscellaneous pH 6.22 6.08 5.98 5.88 Total Suspended Solids (mg/L) 144 175 498 502 Specific Gravity (g/ml) 1.065 1.068 1.077 1.087 Conductivity (micromhos) 133,100 141,500 157,600 173,200 Δ ATP (rlu) – Microbiological Content 1 3 1 1 Microbiological Content low low low low Langelier Saturation Index (LSI) 1.02 0.84 0.72 0.55 Langelier Potential Scaling Mildly Scaling Mildly Scaling Mildly Scaling

6 SPE 125740

Marcellus Shale Post-Frac Flowback Well B Table 2 presents analytical data from the Marcellus Shale Well B flowback waters taken over a period of 45 days. Well

B is located in northern West Virginia. One primary difference in the Well B waters is their significantly higher levels of barium (Ba) and strontium (Sr). Ba levels reached a maximum of 3100 mg/L in Well B vs. 176 mg/L in Well A. Sr levels reached a maximum of 4,310 mg/L in Well B vs. 1837 mg/L in Well A . In Well B, Ba and Sr levels appear to fluctuate more than in the Well A data. The authors have observed Sr values as high as 15,000 mg/L in flowback waters from other Appalachian Basin Marcellus wells. Potassium levels are more elevated in Well A than in Well B. Also present are less common metal ions, including boron, cobalt and lithium. Magnesium and manganese are often found to correlate well with calcium levels.

TABLE 2—MARCELLUS SHALE WELL B FLOWBACK ANALYSIS DATA

Samp Date

TDS Ca (mg/L)

Mg (mg/L)

CaCO3 Na (mg/L)

K (mg/L)

Fe (mg/L)

Ba (mg/L)

Sr (mg/L)

Mn (mg/L)

SO4 (mg/L)

Cl (mg/L)

4/15 pre-frac

224.38 15.00 2.73 49.44 18.00 1.65 0.25 0.23 0.46 0.06 3.00 183.00

4/26 84839 7100 603 23286 22800 326 3.93 2000 1400 6.69 0.00 50600

4/27 89861 7640 651 24952 24300 346 7.80 1990 1510 7.07 8.87 53400

4/27 105169 8490 714 27432 25100 352 9.70 1870 1670 7.44 156 66800

4/28 116266 10500 893 33879 29400 410 35.30 1980 2200 9.10 139 70700

4/29 123902 11700 996 38419 31100 437 16.20 2480 2860 9.50 2.94 74300

4/30 164081 16700 1400 52071 41700 579 23.50 2230 2570 13.00 165 98700

5/1 140169 14000 1150 44358 34300 477 28.70 2290 2590 11.00 22.70 85300

5/2 146539 16700 1380 53473 39400 535 30.20 3000 3380 13.10 0.19 82100

5/3 161636 17100 1410 54446 40400 543 35.20 2950 3280 13.30 4.97 95900

5/4 164902 16700 13000 103026 37000 496 32.90 3850 4310 12.30 1.15 89500

5/28 39706 2920 243 9281 11500 187 26.50 607 472 2.79 47.60 23700

5/29 50019 3140 273 10187 13000 189 31.40 776 568 2.28 619

5/30 94665 9590 906 31580 35300 518 44.70 2500 1800 7.44 199 43800

5/31 93207 7080 672 23552 24000 340 7.65 2090 1370 5.85 41.50 57600

6/1 108047 7860 751 26258 26500 367 7.49 2400 1550 6.08 205 68400

6/8 124303 10200 901 33669 30500 425 34.90 3100 1930 6.79 64.80 Nd

SPE 125740 7

POTW Receiving Water Composition Analyses of Marcellus shale well waters entering POTW for disposal and discharge are presented in Tables 3 and 4. The two different receiving stations are situated in the eastern and western portions of the Marcellus play. The eastern POTW is located in Williamsport, Pennsylvania and discharges into the Susquehanna River. The western POTW is located in Waynesburg, PA and discharges into a tributary of the Monangahela River.

TABLE 2—MARCELLUS SHALE FLOWBACK WATER ANALYSES OF DELIVERIES TO AN EASTERN RECEIVING POTW

Eastern Receiving POTW

1 2 3 4 5 6 7 Avg

TDS 20,90

0 1,850 18,30

0 16,500

38,800

219,000

16,200

47364

CONDUCTIVITY

31,900

3,180 24,800

22.9 37.4 177,000

19.2 33851

OSMITIC PRESSURE

610 58.7 593 460 1,110 7,713 423 1567

SODIUM 5,020 386 4,340 3,660 10,200

69,800 3580 13855

STRONTIUM 22 1.800 20.8 13 60 653 17.3 112 CHLORIDE 12,60

0 815 12,40

0 8,260 21,20

0 169,000

8,080 33194

SULFATES <100 <100 <100 116 205 1,010 63.4 349 IRON-TOTAL 202 0.37 1.44 1.38 2.35 45.5 9.99 38 IRON-DISSOLVED

202 0.03 0.02 <0.10 0.431 0.04 0.04 34

LEAD 0.07 <0.02 <0.02 <0.01 <0.01 <0.02 <0.02

0.07

LITHIUM 0.86 0.1 0.089 0.099 5.09 34.6 0.88 5.96 MAGNESIUM 68.3 7.11 51.9 35.1 87.40 927 50.3 175 MERCURY 0.000

2 0.0002

0.0002

0.0002

0.00 0.004 0.0002

0.00

NICKEL 0.11 0.01 <0.01 <0.12 <0.12 0.16 0.01 0.07 ALKALINITY (T-CACO3)

128 54 172 119 211.00

236 138 151

ACIDITY 314 6 <2.00 130 16.00 528 <2.0 199 ALKALINITY (T-CACO3)

128 54 172 119 211.00

236 138 151

BROMIDE <100 <100 <100 688 25.4 356 CALCIUM 739 176 824 410 917 8,610 574 1750 HARDNESS (TOTALAS CaCO3)

2,500 500 2,400 1,200 2,700 32,400 7,600 70432

BORON <0.01 0.12 0.59 <0.48 3.59 57 0.75 12.41 MANGANESE 20.2 2.42 0.25 3.4 0.606 4.4 0.94 4.60 SELENIUM <0.02 <0.02 <0.02 <0.01 0.029 <0.02 <0.0

2 0.03

Ph 6.17 7.5 10.1 7.29 8.86 6.91 9.14 8.00

8 SPE 125740

TABLE 3—WESTERN RECEIVING POTW REPORTED CHEMICAL ANALYSIS Batch # CHLORIDE - EPA 325.2

1 15,668 2 17,990 3 16,088

4 18,546 5 17,077 6 19,088 7 18,223 8 15,437 9 18,334

10 16,089 11 19,760 12 18,033 13 17,443 14 21,998 15 22,789 16 16,992 17 23,800 18 16,990 19 22,988 20 23,755 21 18,099 22 21,122 23 19,890 24 17,231 25 16,555 26 18,997 27 15,778 28 13,900 29 15,890

Avg 18,433

SPE 125740 9

Geologic and Hydrogeologic Setting Stratigraphic Section In reviewing the hydrogeologic setting, it is important to first examine the stratigraphic section. Pennsylvania is historically one of the most studied stratigraphic sections in the world given the long history of geologic investigations of the region. For this discussion, focus is placed on the overall package of rock deposited during the Middle Paleozoic within the basin and in known evaporite deposits. The stratigraphic section illustrated in Fig. 7 shows the subjacent lithostratigraphic succession on a time-stratigraphic scale.

Fig. 7—Stratigraphic section of major subsurface oil and gas producing zones in Pennsylvania, including the Marcellus Shale (marked with arrow). Harper and Cozart (1992).

Silurian Salina Group The Upper Silurian Salina Group stratigraphically underlies the Devonian Marcellus Shale (Fig. 7). The Salina section is well known for extensive deposition of halite-rich evaporites across the northern Appalachian Basin. Historically, brine extracted from solubilized halite from around Syracuse, NY, near Onondaga Lake, became the most important source of salt in the United States through much of the 19th century (Yager et al. 2007). The halite-dominated intervals grade laterally into anhydrite-bearing shale beds toward the eastern basin margin and anhydrite-dolomite shallowing-upward sequences in the western basin margin (Blauch 1988). Paleogeography of the region is marked by large restricted interior evaporate basins (Fig. 8).

10 SPE 125740

Fig. 8—Early Silurian Paleography. Map after R.C. Blakey. Marcellus Shale Depositional System The Devonian black shale units of the Appalachian Basin, including the Marcellus Shale, are a result of fluctuating depositional settings influenced by sea level fluctuations and plate tectonics-driven basin dynamics. Evidence of basin restriction has been described (Lash 2008). Such fluctuations and basin restriction with periodic replenishment are consistent with generation of deep to shallow subaqueous evaporite conditions, although it is not well accepted that the Marcellus Shale depositional system was deposited as a classic subaerial evaporite sequence (Fig. 9). Deep-water evaporites are well documented in the literature (Matthew and Egleson, 1974) and provide an analog for not excluding evaporite minerals within the deep restricted basin setting, although classification as a true evaporite deposit would not be appropriate for the Marcellus. It is interesting to note a close correlation between the Marcellus lithology and the sulfurous black calcareous shales present in the mid-Michigan Basin Salina A-1 salt. This shale lithotype was deposited under euxinic conditions in a deep basin with restriced circulation, as reported by Matthew and Egleson (1974). It is thought that such black shale deposition occurred during periods when oceanic reflux was severely limited in the Northern Appalachian Basin. Could such hyperpycnal saline conditions have existed during deposition of the Marcellus and other black shales in a silled deep basinal setting? This question requires further investigation. High brine concentrations tend to produce “dead zones” which in turn help to preserve any allochthonous organics. A dense pod or wedge of bottom water also deflects normal bottom-hugging circulation, turning it instead into interflow higher in the water column, further keeping bottom water oxygen levels anoxic. This setting also prevents the heavier bottom brine from being diluted. A one-for-one correlation of the Marcellus to the Michigan Basin Salina A1 depositional setting is not inferred, but restriction and influx of terrigenous material may have obscured the potentially saline brines that may have existed in the Marcellus bottom water setting.

SPE 125740 11

Fig. 9—Paleogeography of the Marcellus Shale Basin. Map after G.G. Lash and R.C. Blakey. Deep Basinal Saline Fluids Within the stratigraphic section of the Appalachian Basin, deep basinal saline brines are well documented and common. Brine salinities within the Lower Devonian Oriskany are shown in Fig. 10. As can be seen, the chemical signature of the pore fluids contained within the Oriskany and other brine-saturated strata are highly saline and in some cases exhibit a chemical signature similar to that of the Marcellus flowback fluids. These brines are hypothesized as being derived from dissolution of the Salina evaporites and emplaced hydrodynamically during deep burial. Salinities in excess of 265,000 mg/L TDS are present within the Oriskany. Since the Oriskany is not considered an evaporite but is rather a group of shallow marine deposits, the presence of hypersaline pore fluids is clear evidence of subsequent migration of evaporite-derived brine into the porous strata. A potential source for this brine is either from other hypersaline fluids or from the dissolution of the subjacent Salina salts or possibly both.

A comprehensive study examining the geochemistry of deep basinal brines in Pennsylvania was conducted by Dresel, 1985. In his study on oilfield brines from Pennsylvania, Dresel concludes that the brines originate as an evaporated seawater that has subsequently been diluted to current concentrations. The brine chemistry has been affected by diagenetic processes such as dolomitization and other water-rock interactions including pyrite oxidation, silicatate alteration, and precipitation of barite, strontianite, celestite, and siderite. Barium concentrations in brine from the Devonian horizons are characteristically high and have potential to form barite deposits. Typically, brines ranged from 9.99 to 343 g/L total dissolved solids. They are characterized by Na-Ca-Cl waters. Their bromide to chloride ratio is greater than that of seawater. Strontium and barium may exceed magnesium concentrations. Iron is high with values reported as high as 494 mg/L and potassium levels ranging up to 4,860 mg/L. The most concentrated samples reported approach the saturation of halite. All samples reported by Dresel (1985) are saturated with respect to barite, calcite and dolomite. Celestite (SrSO4) is close to saturation in his Medina and some Upper Devonian samples, while strontianite (SrCO3) is near saturation in the Oriskany and many Upper Devonian brine samples. Anhydrite (CaSO4) and gypsum (CaSO4·2H2O) also approach saturation in the Medina and some Upper Devonian samples in Dresel’s work.

Over the Marcellus Shale’s regional extent, brines derived from the statigraphic section in porous formations such as the Oriskany are remarkably similar in origin. It is clear that this brine originated as an evaporated seawater, even though the sampled strata that do not contain evaporites themselves. These fluids were mobile at some time in their geologic history and may still be mobile. Evaporite deposits are present in the subsurface Silurian in western Pennsylvania and drilling records suggest that evaporite deposits extend into central and northeastern Pennsylvania.

12 SPE 125740

Fig. 10—Oriskany produced water salinity map. Hypotheses for the Salt Origin in Flowback Waters Primary Dissolution of Autochthonous Salt The term autochthonous applies to “rocks of which the dominant constituents have been formed in situ, e.g. rock salt” (Dictionary of Geological Terms 1976). Primary dissolution of evaporite minerals including halite is perhaps one of the first concepts that come to mind when seeing such high levels of salinity. An excellent example of primary dissolution of autochthonous salt is described by Yager et al. (2007) in the modeling of halite brine in the Onondaga trough near Syracuse, NY. In their study, they observed resultant chloride concentrations ranging from around 20,000 to over 200,000 mg/L.

Evidence of autochthonous salt should be present in the rock matrix within the package of rock contacted by hydraulic fracturing. In order for this hypothesis to test positive, salts and evaporite minerals should be able to be observed in subsurface samples of Marcellus shale, and the cations and anions typical of such evaporite minerals should be present. Primary Dissolution of Allochthonous Salt In the context of this paper, the authors are referring to soluble salts and minerals as “allochthonous” that may have formed from hydrologeologic emplacement and subsequent crystallization in fractures and pores. Confirmation of this hypothesis would be through direct observation of fracture-filling and/or pore-filling evaporite minerals, including halite, from subsurface core. This mode of salt formation would not require a hypersaline or evaporite depositional setting because the origin of the salt would be from outside the rock’s depositional sediment. Re-distribution via dissolution-and-reprecipitation within the same local rock package would also constitute an allocthonous origin, but in a shale should be limited in scope. Encroachment of Basinal Brine This hypothesis for the origin of highly saline flowback fluids, invokes the direct dynamic communication of brine into the hydraulic fracture system, sourced from an adjacent extraneous zone. Saline brine is observed in post-frac flowback fluids typically because of a breach into a more permeable zone that contains mobile brine. A well documented example is found in the Barnett Shale, where the underlying Ellenburger Limestone, when breached, can cause the influx of saline brine. With this mechanism, well-to-well variation in water chemistry is high since only a fraction of the wells breach during fracturing. Mobilization of Hypersaline Connate Fluid This hypothesis involves the mobilization or solubilization of otherwise immobile, highly saline, connate pore fluids held in the subject formation by capillarity or encapsulation. This differs from direct solubilization of a solid-phase salt and requires a water phase to be present in the formation. Combinations Combinations of the above potential hypotheses can be envisioned.

SPE 125740 13

Testing the Hypotheses Compositional Comparisons and Analogs A number of evaporite minerals are most commonly found in modern saline lakes (Fig 11). Evaporite mineralogy of modern salt lake deposits can show appreciable variation, but are easily distinguishable from other non-evaporite settings. Details of the physical chemistry of salt precipitation and earlier work can be found in Stewart (1963); Borchert and Muir (1964); Braitsch (1971).

As an illustration, selected late-stage Marcellus shale flowback waters are shown in comparison to modern day salt lakes (Fig. 11). In a general sense, Marcellus shale flowback waters show some similarity to known, modern day hypersaline evaporites in terms of their aqueous chemistry. An evaporite analog for the original source of ions appears plausible and supported by a number of analogs involving subsurface evaporite systems as well (Dresel, 1985). Is this an indication that the salts are ultimately derived from another formation of a dramatically different origin, coming into the Marcellus via migration of deep evaporite sourced basinal fluids? Could exposure to these fluids occur along routes that may be being opened anew by fracture stimulation? These and other questions can best be addressed through detailed geochemical studies.

Fig. 11—Basic water chemistry comparisons between selected modern salt lakes and Marcellus Shale post-frac flowback waters.

14 SPE 125740

Ca2+

Mg2+

Na+ + K

+CO3

2- + HCO3

-

SO42-

Cl-

SO4

2- + C

l- Ca 2+ + M

g 2+

EXPLANATIONMarcellus Flowback WaterDead Sea GeochemistryOriskany Geochemistry

5638.1341586.6

Early Flowback

Late Flowback

Typical Freshwater

Mixi

ng lin

e

Reused Frac Water

Mixing line

Fig. 12 – Piper diagram of Well A flowback water, Oriskany produced water and Dead Sea water.

In order to further investigate the evaporite source analogy and determine whether Marcellus flowback composition falls

along a dissolution or brine dilution path, analyses of flowback water samples from Well A were plotted on a Piper quadralinear diagram along with data from Oriskany samples from Dresel (1985) and the Dead Sea (Katz and Starinsky, 2008). Piper analysis is a useful tool to interpret mixing versus dissolution routes for waters that have been modified from their original source. Based on the Piper diagram (Fig. 12), if the initial frac fluid was fresh surface water, as is the case for Well A, the data indicate that direct mixing with Oriskany-type brine is not a major factor controlling the observed chemistry of the flowback water from Well A. However, if the initial frac fluid were reused flowback water, the data might suggest a mixing with Oriskany water. Well A is located in a region where subsurface produced brines in subjacent strata are not known to be present.

The Dead sea sample is characterized as calcium-magnesium-chloride type while the Oriskany sample is sodium-chloride type. All of the tested flowback waters can be characterized as belonging to the sodium-chloride facies, however, the water clearly evolves from being moderately sodium-chloride type to a more strongly sodium-chloride facies, similar in composition to the Oriskany samples. The flowback water shares no genetic characteristics with the modern Dead Sea type water other than similar sodium and chloride levels and an evaporative source. Freshwater is prototypically characterized as a weak calcium-magnesium-bicarbonate type. Since Well A was stimulated using surface freshwater, the data indicate that mixing is not the main controlling factor on the flowback water geochemistry, but rather perhaps dissolution of preexisting salts and minerals is the controlling factor. Operators are known to reuse flowback water during fracs. If that mixing path were followed, the data would suggest that the flowback water chemistry is controlled by mixing between the frac fluid and Oriskany water, as opposed to dissolution of Marcellus salts. Analysis of Subsurface Core for Residual Salts One of the key concepts for investigation of the salt question and further testing the Piper analysis findings above is to look specifically at Marcellus shale rock in the form of both core and drilled cuttings taken from the subsurface to rule in or out the possibility for indigenous autochthonous vs. extraneous allochthonous salt. Although not previously described in literature on the Marcellus, minor volumes of dispersed salts may be present that are visually obvious from routine core

SPE 125740 15

examination. Mineralogic and petrographic analysis can provide critical insight into this issue. During hydraulic fracturing of shale, extreme volumes of rock surface area are contacted. It is plausible that even trace levels of salts contained within the shale matrix may contribute to saline fluids because of the very high rock/fluid volume ratio. Localized concentrations of soluble salts and minerals could also be present as microscopic dispersed grains, laminae or nodular blebs rather than as bedded deposits typical of conventional evaporites. Allocthonous salts and minerals could occur within fracture networks or previously open pores in lithotypes containing high carbonate content shales, concretions or limestone beds. Mineralogy and Petrography of Selected Marcellus Lithotypes Samples were examined from selected drill cuttings obtained from a Marcellus shale well. Thin sections of all samples were prepared by following the industry-standard methods of thin section preparation, e.g. (1) epoxy impregnation in vacuum and curing, (2) sectioning, (3) grinding, (4) precision sectioning, and (5) precision grinding. For XRD and chemical analyses, samples were pulverized in a micronizing mill down to less than 10-microns size. XRD analyses were carried out by using a Siemens D 3000 Diffractometer, using theta-two theta geometry. Samples were scanned at 2 degrees/min rate, at 30mA current and 40kv voltage, using copper Kα radiation. The diffraction patterns were searched by using the MDI-Jade 8.0 software with search-match module. Chemical analyses were done by using inductively coupled plasma-atomic emission spectroscopy (ICPAES) for most of the major and trace elements. The EPA methods were used for Cl, F, and sulfate analyses. Combustion Infra Red (IR) was used for determination of carbon and sulfur contents, and methods of ASTM D 2216 were followed for determination of moisture content. Additionally, hydrochloric acid-insoluble contents were measured to determine the variation in carbonate contents. Scanning electron microscopical examinations with ancillary energy-dispersive x-ray microanalyses were carried out by using a Cambridge (Camscan) Series II scanning electron microscope (SEM) equipped with Robinson backscatter detector, SE-detector, and 4Pi-EDS analysis. Summary of Results XRD patterns showed the presence of major amounts of quartz and clay minerals, and minor amounts of calcite, halite, pyrite, and mica in the argillaceous 5145-5160 sample, whereas major amounts of calcite and minor amounts of quartz showed in the 5220–5236 calcareous-argillaceous sample.

Petrographic analysis explains the observed compositional differences (Figs. 13 and 14). Sample 5160–5175 showed micro-fossiliferous shale but also carbonate-mottled, claystone lithotypes, suggesting a possible evaporite type morphology and texture. Sample 5220–5235 showed micro-fossiliferous shales with some lithotypes containing well-defined geopetal fabric and microcrystalline infill.

Hydrochloric acid-insoluble residue (IR) contents were 93.1 % for shale sample 5145–5160, but only 16.8 % for the carbonate-plus-shale sample 5220–5235.

The major element oxide compositions were typical of argillaceous rocks in the 5145–5160 sample, whereas an indication of a mixture of major amounts of carbonate and subordinate amounts of shale showed in the 5220–5235 sample.

Treatment with dilute hydrochloric acid showed strong effervescence in the carbonate-based 5220-5235 sample, but negligible to no effervescence in the argillaceous-based 5145-5160 sample.

Halite was present as a mineral phase, although as a subordinate phase as determined by XRD analysis (Fig 15). The major differences in chemical compositions of two samples can be summarized as shown in Table 5.

TABLE 4—ELEMENTAL ANALYSIS OF SELECTED MARCELLUS SHALE DRILLED CUTTINGS

Sample SiO2+Al2O3+Fe2O3

Na2O+K2O CaO S C LOI IR

5145-5160 73.3 4.56 1.75 4.4 7.3 0 93.1

5220-5235 16.3 0.76 44.8 0.46 10 24 16.8

16 SPE 125740

Fig. 13—Thin-section photomicrograph of sample 5160–5175 showing micro-fossiliferous shale lithotypes and carbonate-rich

mottled lithotype suggesting a possible evaporite morphology and texture.

Fig. 14—Thin-section photomicrograph of sample 5220–5235, showing micro-fossiliferous shale lithotype containing well defined geopetal fabric and microcrystalline infill.

SPE 125740 17

Fig. 15— XRD analysis of Marcellus shale sample 5220–5235 showing the presence of Halite along with the presence of major amounts of calcite, and subordinate amounts of quartz.

Salt Dissolution Studies from Whole Core A section of Marcellus core was selected based on visual observation of what appeared to be salt precipitation along bedding laminae or parting planes (Fig 16). The primary question to be answered is “Are the observed layers of autochthonous salt deposited with the shale or are they artifacts of the coring process?” To answer this question, scrapings of the “salt” layers were taken, along with a section of the bulk core between the “salt” layers. Analytical methods applied to these included dissolution in a weak 1% nitric acid solution to facilitate solution of cations that are associated with salts such as halite or other evaporates such as anhydrite, kieserite (MgSO4·H2O), celestite, polyhalite (K2Ca2Mg(SO4)4·2H2O), or sylvite (KCl). Primary cation and anion analysis is used to compare the observed “salt” layers with the bulk core material, which was crushed to increase surface area. The samples were each digested and analyzed using ICP and bulk anion analysis.

18 SPE 125740

Fig. 16 —Slabbed Marcellus shale core section containing salt layers (light areas) along bedding plane.

Core vs Salt Scraping Results Using the absolute values reported, the core sample exhibits ten times the iron, three times the Mg and similar Ca and K contents of the scrapings. The bulk core exhibits approximately 1/3rd of the Ba, 1/20th of the Na, and 1/40th of the chloride content as the scraping sample. From the results of the analyses shown in Table 6, it can be concluded that both the bulk core and the scrapings are consistent with natural autochthonous salt obtained along bedding planes. Key elements observed in both the bulk core and scraping of the salt layers are common to both the flowback fluids and the elements observed dissolved from the core. Common cations include Ba, Sr, Ca, Na, Fe, Mg, and K. These are not consistent precipitants one would expect from the mud used in coring the well nor filtrate from that mud.

TABLE 6—RELATIVE SOLUBLE CATION CONTENT

Cation Bulk Core Salt Scraping Ba 1% 2% Ca 40% 42% Fe 46% 5% K 4% 7% Mg 6% 3% Na 2% 40% Sr 1% 1% Total 100% 100%

Implications Analysis of the flowback geochemisty can provide important clues regarding source and origin of ions picked up during the fracturing process. Deposition environments interpreted for the Marcellus shale may or may not have a direct impact on the flowback geochemistry. If salts observed in the Marcellus were deposited in a restricted basinal system, these findings it may have implications to other shales. If observed ions found in the Marcellus shale samples are syndepositional autochthonous, then some re-interpretation of the deposition setting is needed, as least in part. If the observed ions found in the Marcellus shales are allochthonous, based on direct textural evidence we have yet to gather at this writing, then the migration of deep basinal brine into the shale package over geologic time must be inferred. Obviously, there is still more work to be done in this area. Evaporite type fluids appear ubiquitous in the sedimentary package found subjacent to the Marcellus shale in porous Devonian sandstone units such as the Oriskany and Medina. While the analyses presented here do not identify the flowback waters to be a mixing of waters from these porous subjacent units, neither can we completely rule out the possible contribution of such a source conclusively. Obviously, there is still more work to be done in this area as well.

SPE 125740 19

Elevated concentrations of barium and strontium in addition to high salinity observed in Marcellus shale flowbacks may directly impact pollution potential of discharged brines. The level of TDS in them make treatment difficult and expensive and with the implementation of planned new threshold limits on discharges in Pennsylvania may well inhibit continued development of the Marcellus in the way that it has been done in the past. Reuse of fracturing fluids would likely impact the geochemistry of resultant flowback chemistry. One implication for water reuse is the potential for reduced levels of solubility and therefore less saline burden to the environment over time.

Problems associated with fluids management after a large slickwater treatment are serious, no matter the origin of the brine, but are particularily a problem when the brine density increases from high ionic strength. Failure to remove fluids from the wellbore can lead to lower production rates due to an unfavorable flow regime like bubble flow, and can reduce the effective fracture half-length and production by back filling the gas saturated flow paths with brine. The implications are for both vertical and horizontal wells.

Based on data presented in this paper, reuse of the flowback water directly or when dilute with fresh water will have a direct impact on fracturing chemical selection. Anionic fresh water friction reducers lose performance with increasing salinity and hardness. There are currently no industry established cutoff points for TDS, hardness, and divalent metal ions for friction reducer use. Research and development has led to the introduction of a limited number of higher TDS and divalent ion friction reducers but prudent operators should check with their pumping service providers for the latest information before pumping flowback water mixtures into any shale. Pre-job friction loop performance testing should always be performed before each job to ensure proper mix ratio of reuse water and to ensure proper functional performance of the friction reducer in the presence of the actual waters being pumped.

Recycling of flowback water into fresh water using reverse osmosis, distillation, or EC is expensive and directly impacted by the water chemistry of the flowback waters. Another important implication of this work is potential scale formed insitu by pumping fresh water with appreciable quantities of ionic species such as sulfate and mixing with calcium, barium, and strontium rich connate water. Geochemical modelling should be conducted and appropriate mitigation steps taken to reduce the risk.

An implication of high salinity late stage flowback fluids is deposition of salt similar to the problem experienced by Oriskany operators. Salt deposits may show up in surface lines where a pressure drop occurs in the more saline late-stage flowback waters.

Divalent cations at levels observed impact the success of water reclamation treating technology and efforts to reuse the flowback water as a fracturing fluid. Future Work By intent, this paper only introduces the reader to this important and complex subject. Further work is needed in many areas. The type of work presented here should be expanded. Isotopic studies should be carried out to provide further evidence as to the precise origin and distribution of constituents observed in Marcellus shale post-frac flowback fluids. Further ionic distribution mapping could lead to key correlations across the basin and could aid in determination of production sweet spots. The authors are currently in the process of gathering additional data and conducting specific ionic ratio analysis across the basin. Further analysis of core and cuttings is ongoing based on results reported herein. Similar work by others will welcomed. Future water reuse initiatives should be studied with respect to fluid-shale interaction and geochemical equilibration. Conclusions Gathering of well post-frac flowback geochemistry data provides useful information when integrated with geologic, mineralogic, and basin-wide production water data regarding the source and mechanism of observed soluble ions found in solution. Several hypotheses are presented for explaining the origin of salts and dissolved ions observed in Marcellus shale flowback fluids. Interpretation of the Marcellus shale depositional system may indicate potential for primary restricted deep basin saline brine during periods of basin restriction. However, this issue needs further investigation across the basin for evidence of syndepositional evaporative minerals versus allocthonous emplacement from basinal evaporite derived brines during deep burial diagenesis. A correlation exists between the evaporite signatures of associated produced brines and post-frac Marcellus shale flowback fluids. Direct observation of halite is observed in Marcellus shale core both within the bulk matrix and inconcentrations along horizontal bedding planes in discontinuous lenses. Marcellus shale contains the metal ions including Na, Ca, Mg, K, Fe, Sr, and Ba that are observed in flowback fluids. Piper diagram analysis indicates primary dissolution as a likely source of dissolved ions in at least one well studied in this paper. All flowback water presented in this paper is characterized as belonging to the sodium-chloride facies. This water signature clearly evolves from being moderately sodium-chloride type to a strongly sodium-chloride facies, similar in composition to the Oriskany samples. Key cations and anions such as Sr, Ba, K, Li, Mg, Mn among others can provide important chemical fingerprints for basin wide geochemical studies.

20 SPE 125740

Although local variations in flowback chemistry exists, at the regional scale, the degree of homogeneity and consistency in the flowback chemistry suggests a large scale, basin wide control rather than local controls such as natural fractures. Reuse of fracturing fluids is an important means to avoid surface discharge. However, geochemical modeling and understanding of the implications of water reuse on the geochemical equilibrium are needed. Acknowledgements The authors would like to thank their respective companies, EXCO-North Coast Energy, Inc. and Superior Well Services, Inc. for their support in preparation of this paper and their permission to publish. A special thanks to Paul Rey and Dave Grottenthaler for their analytical work and assistance in gathering of water chemistry data. References Blauch, M.E. 1988. Dolomitization, diagenesis, and paleoenvironments of a middle to upper Silurian shallowing upward sequence exposed

near Peebles, Ohio. MS dissertation, University of Akron, Akron, Ohio. Blakey,R.C., 2006, Paleogeographic Maps Arranged as Time Slices in www2.nau.edu/rcb7/global_history.html. Borchert, H. and Muir, R.O. 1964. Salt deposits: the origin, metamorphism and deformation of evaporates. Van Nostrand. London. 338. Braitsch, O. 1971. Salt deposits, their origins and composition. Springer. Berlin. 297. Dean, W.E. and Schreiber, B.C. (Eds.). 1978. Notes for a short course on marine evaporates. Soc. econ. Paleon. Min. Tulsa, Short Course. Dictionary of Geological Terms, Revised Edition. 1976. Garden City, New York: Anchor Press/Doubleday. Dresel, P.E. 1985. The Geochemistry of Oilfield Brines From Western Pennsylvania. MS Thesis, The Pennsylvania State University

Department of Geochemistry and Mineralogy. Effler, S. W., Doerr, S.M., Brooks, C.M., and Rowell, H.C. 1990. Chloride in the Pore Water and Water Column of Onondaga Lake, N.Y.,

U.S.A.: Water, Air, & Soil Pollution. 51, 315–326. Halite Brine in the Trough near Syracuse. 2007. Scientific Investigations Report. US Geological Survey. 5058. Yager, Kappel, and

Plummer. New York. Horita, Juske, Wesolowski, D.J., and Cole. D.R. 1993.The activity-composition relationship of oxygen and hydrogen isotopes in aqueous

salt solutions: I. Vapor-liquid equilibration of single salt solutions from 50 to 100°C. Geochimica et Cosmochimica Acta. 57. 2797–2817.

Kappel, W.M. 2000. Salt production in Syracuse, New York (“The Salt City”) and the hydrogeology of the Onandaga Creek Valley: U.S. Geological Survey Fact Sheet 130–00.

Katz, Amitai, and Abraham Starinsky. 2008. Geochemical History of the Dead Sea. Aquatic Geochemistry. Vol 15, pp 159-194. Kendall, A.C. 1979. Continental and supratidal (sabkha) evaporates. Subaqueous evaporates. In: Facies Models (Ed. By R.G. Walker).

Geoscience Canada. 145–174. Kirkland, D.W. and Evans, R. (Eds.). 1973. Marine evaporates: origins, diagenesis, and geochemistry. Dowden, Hutchinson, and Ross.

Stroudsburg. 426. Krauskopf, K.B., 1979. Introduction to Geochemistry, pp. 617,. McGraw-Hill, New York. Kurlansky Mark. 2003. Salt—A World history: New York. Penguin Books. Lash, Gary G. 2009. Overview of the Tectonic and Eustatic Framework of Yager, Kappel, and Plummer. Devonian Black Shale,

Appalachian Basin. Presented at the SPE Cooperstown Applied Technology Workshop, Cooperstown, New York. Matthews, R. D. and Egleson, G. C. 1974. Origin and Implications of a Mid-Basin Potash Facies in the Salina Salt of Michigan, Presented

at the Fourth International Symposium on Salt – Northern Ohio Geological Society. Oil and Gas Developments in Pennsylvania in 1990 With Ten-year Review and Forecast. 1992. Progress Report 204, Commonwealth of

Pennsylvania. Harper, J.A. and Cozart, C.L. Pennsylvania. Rickard, L.V. 1969. Stratigraphy of the upper Silurian Salina Group New York, Pennsylvania, Ohio, Ontario. New York State Museum and

Science Service Map and Chart Series No. 12. Tucker, M.E. 2001. Sedimentary Petrology An Introduction. Geoscience Texts Vol. 3. A Halstead Press Book. New York, Toronto. John

Wiley & Sons.