- Organic matter in flowback water is dominated by low molecular weight fractions.
- Low molecular weight organics are effectively removed via biodegradation.
- Organics with high molecular weight are targeted by activated carbon.
- Biodegradation and activated carbon may complement flowback desalination.
Andrii Butkovskyia, Ann-H-el_ene Faberb, c, Yue Wanga, Katja Grollea,Roberta Hofman-Carisb, Harry Bruninga, Annemarie P. Van Wezelb,c, Huub H.M. Rijnaartsa
aDepartment of Environmental Technology, Wageningen University, P.O. Box 17, 6700 AA, Wageningen, The Netherlands. bKWR Watercycle Research Institute, P.O. Box 1072, 3430 BB, Nieuwegein, The Netherlands. cCopernicus Institute of Sustainable Development, Utrecht University, Heidelberglaan 2, 3584 CS, Utrecht, The Netherlands
Received 8 January 2018 Accepted 15 March 2018 Available online 15 March 2018
Ozonation, sorption to granular activated carbon and aerobic degradation were compared as potential treatment methods for removal of dissolved organic carbon (DOC) fractions and selected organic compounds from shale gas flowback water after pre-treatment in dissolved air flotation unit. Flowback water was characterised by high chemical oxygen demand and DOC. Low molecular weight (LMW) acids and neutral compounds were the most abundant organic fractions, corresponding to 47% and 35% of DOC respectively.
Ozonation did not change distribution of organic carbon fractions and concentrations of detected individual organic compounds significantly. Sorption to activated carbon targeted removal of individual organic compounds with molecular weight >115 Da, whereas LMW compounds remained largely unaffected. Aerobic degradation was responsible for removal of LMW compounds and partial ammonium removal, whereas formation of intermediates with molecular weight of 200–350 Da was observed. Combination of aerobic degradation for LMW organics removal with adsorption to activated carbon for removal of non-biodegradable organics is proposed to be implemented between pre-treatment (dissolved air floatation) and desalination (thermal or membrane desalination) steps.
The negative impact on the quality of surface and ground water is one of the major environmental consequences of shale gas production (Brantley et al., 2014; Ferrar et al., 2013; Jackson et al., 2013; Osborn et al., 2011; Vidic et al., 2013). Shale gas producers use a combination of horizontal drilling and hydraulic fracturing to recover gas from formations. Hydraulic fracturing implies high pressure well injection of large volumes of water mixed with inert solid material (proppant) and chemical additives.
This mixture, called fracturing fluid, creates fissures in the shale increasing its permeability and gas recovery. The mixture of the injected fracturing fluid together with the connate water of the formation, which returns to the surface within few weeks after pressure being released, is called flowback water (Olmstead et al., 2013). The connate water of the formation, which continues flowing upwards for years after fracturing has been completed, is called produced water (Vidic et al., 2013).
Both flowback and produced waters (FPW) are contaminated with high concentrations of total dissolved solids (TDS), oil and grease, natural radioactive materials (NORM), and dissolved organic matter (DOM) (Gregory et al., 2011). Injection into disposal wells was the most widely used shale gas wastewater management strategy until 2010 (EPA, 2016; Rahm et al., 2013). However, the limited availability of disposal wells, increased uncertainties about risks related to deep well injection, and legislative constrains make the industry turn towards reuse and discharge of FPW (Gregory et al., 2011; Mauter et al., 2014; Rahm et al., 2013; Silva et al., 2017). Removal of contaminants is required before FPW can be discharged.
FPW reuse also requires removal of certain compounds, e.g. potential scalants and foulants. Treatment technologies typically include separation for removal of total suspended solids (TSS), oil and grease, adsorption for removal of organics, NORM and heavy metals, membrane or thermal desalination for TDS removal (Drioli et al., 2015; Igunnu and Chen, 2014; Jiménez et al., 2018; Saba, 2014). Desalination is a crucial step in the FPW treatment, because of the high TDS concentrations that often equal or exceed sea water salinity (Shaffer et al., 2013). Membrane and thermal desalination technologies for oil and gas produced water are well established and can remove TDS and majority of contaminants mentioned above (Abousnina et al., 2015; Fakhru’l-Razi et al., 2009).
However, concentrations of organic carbon in shale gas FPW often exceed 1000 mg/L, causing fouling on the membranes and distillation equipment (Alzahrani and Mohammad, 2014; Chen et al., 2015; Thiel and Lienhard, 2014). In addition, desalination technologies are effective towards high molecular weight polar compounds, such as polycyclic aromatic hydrocarbons (PAHs), but poorly remove low molecular weight non-polar organics, that may also pose threats to the environment and human health (Annevelink et al., 2016; Butkovskyi et al., 2017; Ferrar et al., 2013; Shaffer et al., 2013). Despite that only few studies attempt to characterize the nature of organic compounds in FPW and evaluate potential treatment strategies for their removal (Butkovskyi et al., 2017; Camarillo et al., 2016).
Components of the fracturing fluid, including heterocyclic biocides, cocamidopropyl surfactants, ethylene glycol and derivatives, as well as natural constituents of shale, including aliphatic and aromatic hydrocarbons, alkanes, resins, asphaltenes, heterocyclic and halogenated organics were detected in FPW from U.S. shale basins using gas and liquid chromatography coupled to mass spectrometry (GC-MS and LC-MS) (Annevelink et al., 2016; Butkovskyi et al., 2017; Ferrar et al., 2013; Hayes, 2009; Lester et al., 2015; Maguire-Boyle and Barron, 2014; Orem et al., 2014).
Lester et al. (2015) reported low removal of DOM from shale gas flowback water by advanced oxidation, whereas aerobic treatment removed >50% of DOM at flowback water TDS of 22.5 g/L. Other authors demonstrated that microbial mats are capable to decrease COD at TDS as high as 100 g/L (Akyon et al., 2015). Neither the fate of different DOM fractions nor the removal of individual organic compounds were studied in these works, moreover, Akyon et al. (2015) did their experiments with synthetic flowback water. Several authors reported high biodegradability of fracturing fluid components, including poly(ethylene glycol) surfactants and biocide glutaraldehyde (Kekacs et al., 2015; McLaughlin et al., 2016; Mouser et al., 2016). However, the degradation of compounds was studied in fracturing fluids, which composition is very different from FPW.
High potential of granular activated carbon (GAC) for adsorption of fracturing fluid chemicals furfural and 2-butoxyethanol and powdered activated carbon (PAC) for adsorption of polyethylene glycols was also shown (Manz et al., 2016; Rosenblum et al., 2016). So far these are the only two studies focused on application of activated carbon for removal of organic chemicals from flowback water.
This paper aims to assess the removal of different fractions of dissolved organic carbon and individual organic compounds present in flowback water using typical primary treatment process (dissolved air flotation) followed by one of the common processes for organics removal (ozonation, adsorption to activated carbon or aerobic degradation). Flowback water is studied because, in comparison to produced water, this stream has typically high concentrations of organic carbon and contains potentially harmful components of fracturing fluids. The most promising treatment strategy for organic compounds removal is proposed and its integration with existing treatment processes is discussed.
Materials and methods
The shale gas flowback water was obtained from a Baltic shale gas basin (Poland) under non-disclosure agreement regarding location, storage conditions and composition of the fracturing fluid used at the production site. The water was sampled within two months after commencement of the first fracturing operation. It was transported in 20 L plastic containers and stored at 4 °C until the experiments. The flowback water was pre-treated in a dissolved air flotation (DAF) unit assisted by coagulation in order to remove TSS, oil and grease. DAF-treated flowback water was used for ozonation, adsorption to GAC and aerobic degradation experiments.
Dissolved air flotation/coagulation
Flowback water (V = 80 L) was treated by FeCl3-assisted DAF in a stainless steel tank (V = 200 L). FeCl3 (100 mg Fe/L) was added as a coagulant and pH was adjusted to 8.0 by addition of NaOH (0.5 g/L) (Megid et al., 2014). Flocculation was observed within 40 min and 3 L of water saturated with air was released to the stainless steel vessel from the adjacent pressurized tank (p = 5 bar). Flotation contact time was 10 min. DAF-treated flowback water (V ≈ 63 L) was collected excluding precipitate and scum (V ≈ 20 L) and stored at 4 °C until ozonation, GAC sorption and aerobic degradation tests were performed.
Ozonation was performed in batch mode in a glass vessel (Vliquid = 2 L) with ozone supplied through a bubble diffuser during 60 min. Ozone was produced from pure oxygen using a Fischer 503 ozone generator. The gas flow was maintained at 0.1 m3/h, and the ozone concentration in the reactor inlet at 6.3 g/m3. Ozone concentrations in the gas phase of the reactor inlet and outlet were measured by BMT ozone analyser 961TC and 961, respectively.
An ozone dosage of 0.3 g/L was applied to DAF-treated flowback water.
Additionally, control tests with air supplied instead of ozone at the same flow rate were run to correct for volatilization. To correct for ozone losses in the system, a blank test with milliQ was run until stabilization of ozone concentrations in the reactor inlet (≈6.0 g/m3) and outlet (≈5.5 g/m3) has been reached. The ozone losses were taken into account when calculating ozone consumption of the flowback water. Liquid samples were filtrated through 0.45 μm pore size filters directly after the test and stored at −20 °C until the analyses.
Sorption to GAC
Sorption tests were performed with DAF-treated flowback water and three different types of GAC. Granular types of activated carbon were chosen because they are preferred to powdered by oil and gas industry due to lower carbon usage rates and operational costs (Arthur et al., 2005; Hackney and Wiesner, 1996).
The tests were performed in stirred glass vessels (Vliquid = 1 L) using three different fresh GAC types typically used in municipal and industrial wastewater treatment, Chemviron F400, Norit GAC 830W and Norit C GRAN (Table S1) and dosage of 2000 mg/L. The adsorption tests were run for 6 weeks at 20 °C to reach the equilibrium state. Liquid samples were filtrated through 0.45 μm pore size filters directly after the test and stored at −20 °C until the analyses.
Batch aerobic degradation test was adapted from OECD method 301A (OECD, 1992). The test was performed in triplicate in 250 mL glass bottles holding a liquid volume of 120 mL and headspace volume of 130 mL. Activated sludge adapted to high salinities was obtained from a wastewater treatment plant (WWTP) at Delfzijl (the Netherlands) which treats industrial wastewater with high Cl− concentrations (2–20 g Cl−/L) (van der Marel and de Boks, 2014).
The sludge (volatile suspended solids (VSS) = 3.5 g/L) was collected from the aerobic nitrification basin and stored at 4 °C until the experiment. Sludge liquor had low COD (170 mg/L), no ammonium nitrogen and comparatively high chloride concentrations (Cl− = 22 g/L).
DAF-treated flowback water was mixed with sludge from Delfzijl WWTP (volumetric ratio of 0.29:0.71) to obtain the sludge loading of 0.1 g COD/g VSS*d. Diluted batch samples prepared for aerobic degradation experiments contained significantly lower COD (610 mg/L) and DOC (192 mg/L) when compared to the DAF-treated flowback water. Disodium phosphate was added as external phosphorous source to reach substrate N:P ratio of 5:1.
The rest of the nutrients required according to OECD method 301A were present in the tested flowback water in excess. The glass bottles were closed with butyl rubber septa and incubated horizontally at constant temperature (20 °C) for 48 h on a linear shaker at 160 rpm for maximum oxygen up-take. The headspace composition was refreshed as soon as the oxygen volume fraction dropped to 10%.
To adapt the sludge to changes in salinity and substrate composition, batch incubations were repeated six times with the substrate refreshment between incubations. To refresh the substrate, the batches were centrifuged at 1500 rpm and the sludge pellet was mixed with a new portion of DAF-treated flowback water and sludge liquor. Full sampling campaign and analytical measurements were performed during the last incubation.
The liquid samples and the headspace gas samples were taken at t = 0.5, 1, 2, 3, 5, 7, 10, 24 and 48 h. Headspace gas composition was analysed immediately after sampling. All liquid samples (V = 1.5 mL) were centrifuged at 10,000 rpm and the supernatant was stored at −20 °C until the analyses.
An abiotic control experiment was performed using DAF-treated flowback water and sodium azide (6.4 mg/L). Control batches were incubated and sampled similarly to the test batches.
Chemical oxygen demand (COD) and ammonium nitrogen (NH4-N) were analysed with Hach test kits LCK-1414 and LCK-304 respectively. The pH was measured with Hach HQ 440d multimeter. VSS of sludge in aerobic degradation experiments were measured according to the Standard method 2540 D (APHA, 1998). Anions were analysed by ion chromatography with conductivity detector (IC-CD; Dionex ICS 2100). Cations were analysed by inductively coupled plasma optical emission spectrometry (ICP-OES; Agilent Vista-MFX) after acidification with 1% nitric acid.
Volatile fatty acids (VFA) and alcohols were analysed by gas chromatography with flame ionization detector (GC-FID; Agilent 7890B) after sample acidification with formic acid. All samples were diluted with milliQ if the concentrations of analytes were above the calibration limits or salt concentration interfered with analytical procedures. The quantification limits for IC-CD, ICP-OES and GC-FID are presented in Tables S2 and S3.
Headspace gas composition in batch tests was analysed by gas chromatography with micro-thermal conductivity detector (GC-μTCD; Shimadzu GC-2010). The concentration of oxygen in the headspace during biological batch tests was quantified according to ideal gas law with pressure inside the bottle measured by pressure meter (Greisinger GMH 3151). Specific oxygen uptake rate (SOUR) was calculated as a ratio of the measured oxygen consumption rate at the defined time period and sludge VSS.
The fractions of organic carbon were separated using size-exclusion chromatography (SEC) followed by multidetection of organic carbon (OCD), UV-absorbing aromatic and unsaturated structures at 254 nm (UVD) and organic bound nitrogen (OND; LC-OCD analysis at the DOC-Labor Dr. Huber laboratory). The detailed description of the analytical equipment and modified procedure for liquids with high TDS are found in the literature (Huber et al., 2011; Salinas Rodriguez, 2011).
Calibration of molecular masses and detector sensitivities was performed with a NaCl-solutions of humic and fulvic acids and potassium hydrogen phthalate. Chromatograms were processed on the basis of area integration using the program ChromCALC. The assignation of the DOC fractions detected in the flowback water is given in Supplementary material.
Organic compounds were analysed using Liquid Chromatography coupled to a Linear Ion Trap Orbitrap High Resolution Mass Spectrometer (LC-LTQ/HRMS) in positive and negative ionisation mode (ses Supplementary material for additional methodological details) (Sjerps et al., 2016). Interpretation of detected peaks was performed using Sieve 2.2 (peak integration) in combination with Xcalibur software (molecular formula identification).
The semi-quantitative concentrations of the attributed compounds were expressed as atrazine-d5 equivalents (IS-eq) in positive ionization mode and as bentazone-d6 equivalents in negative ionization mode with a detection limit of 0.05 μg/L IS-eq. Confidence levels regarding the identification of compounds were reported according to (Schymanski et al., 2014). A level 5 confidence level suggests that an exact mass has been detected and an unequivocal molecular formula based on isotope and adduct information provides a level 4 identification. Level 3 suggests that a number of tentative structures are identified based on MS2 data.
Levels 2 and 1 suggest that a probable structure or a confirmed structure has been assigned based on matches to library spectrums or reference standards. Total IS-eq concentrations of organic compounds were reported with a detection limit of 0.01 μg/L. Carbon, hydrogen, oxygen, nitrogen and phosphorus were allowed for molecular formula determination and, if suggested by the spectrum, chlorine and/or sulphur were also considered.
The direction (+or -) of the internal standard mass error for a given spectrum in combination with isotope information (i.e. presence of sulphur, halogens, the number of carbons based on C13 information, etc.) was used to determine the most likely molecular formula. In addition, a suspect screening was also performed, using the Compound Discoverer 2.0 software (Thermo Scientific). The used suspect list includes chemicals that could potentially be present in shale-gas related waters and is described in detail in (Faber et al., 2017).
Results and discussion
Flowback water composition
The raw flowback water had a low pH (4.9) and high salinity (TDS = 103 g/L) with chloride being the dominating anion and sodium, calcium, magnesium and strontium – the dominating cations (Table S2). The relatively high concentrations of scale-forming cations (Ca, Mg, Sr and Ba) should be of concern when targeting flowback water reuse in hydraulic fracturing because of the high scaling potential (Thiel and Lienhard, 2014). Heavy metals were not detected in the flowback water, except for manganese, which concentrations were in the same order of magnitude as reported in the other studies (Hayes, 2009; Lester et al., 2015; Thacker et al., 2015; Ziemkiewicz and Thomas He, 2015).
The flowback water was also characterised by high COD (1800 mg/L), DOC (649 mg/L) and ammonium nitrogen (103.5 mg/L) concentrations. Fractionation of organic matter according to the size and hydrophobicity with subsequent detection of organic carbon, UV-absorbance at 254 nm and nitrogen showed that LMW acids and neutrals were two dominant organic fractions in the flowback water (Fig. 1; Table S4).
Since UVD and OND detectors showed no response for elution of LMW neutrals, it was concluded that the fraction is represented by saturated hydrophilic compounds which do not contain nitrogen. GC-FID detection of VFA and alcohols has shown that LMW acid fraction was dominated by acetic acid (319 ± 19 mg/L, or 43%) and LMW neutral fraction – by ethanol (215 ± 17 mg/L, or 52%) (Fig. 1; Table S3). Other fatty acids (propionic, butyric and hexanoic acid) and alcohols (propanol and butanol) were also detected, though in much lower concentrations (Table S3).
Significant parts of LMW acids (48.6%) and LMW neutrals (40.5%) fractions were not classified further by GC-FID, suggesting the presence of a mixture of various LMW compounds apart from acetic acid and ethanol. Other DOC fractions included hydrophobic organic compounds (HOC), high molecular weight biopolymers (Mw > 10000 Da) and fraction of building blocks (Mw = 300–500 Da) (see Supplementary material for detailed description).
These fractions presumably contain natural hydrocarbons of the formation and synthetic organic chemicals added to the fracturing fluid. However, the response of LC-OCD system to synthetic organic chemicals used in hydraulic fracturing was not tested since the composition of fracturing fluid used at the flowback water sampling site was not disclosed.
Fig. 1. Fractions of TOC (A) and molecular composition of LMW acid (B) and LMW neutral (C) fractions of the shale gas flowback water.
The concentration of acetic acid in this study (319 ± 19 mg/L) is in the range of the concentrations of acetate measured by Olsson et al. (2013) in flowback water from two hydraulically fractured wells in Germany (417 and 197 mg/L respectively). The same authors did not find acetate in a well which was not fractured, and concluded that acetate is a degradation product of polymers used in hydraulic fracturing fluids (Olsson et al., 2013).
These results are also confirmed by other authors who observed release of organic acids from shale being in contact with fracturing fluid under high temperature and pressure conditions (Vieth-Hillebrand et al., 2017). Contrary, Orem et al. (2014) did not find significant concentrations of acetate in flowback water from hydraulically fractured wells in Marcellus and New Albany shales (USA).
They suggest that acetate, even if formed downhole, will be immediately consumed by populations of methanogenic organisms in the well. According to these authors high concentrations of acetate in flowback water found at some fields can be explained by cracking of kerogens at temperatures exceeding 80 °C, which also inhibits methanogens and thus the conversion of acetate to methane (Orem et al., 2014).
Presence of ethanol in shale gas flowback water was previously reported only in one study, where it was detected in one well out of nineteen and originated from fracturing fluid used at that location (Hayes, 2009). Ethanol may also be a degradation product of long-chain ethoxylated alcohols, which are frequently used in hydraulic fracturing operations.
Twenty-seven compounds with unique m/z ratios were detected in flowback water with LC-LTQ/HRMS at IS-equivalent concentrations ≥0.05 μg/L using Sieve 2.2 and Xcalibur software (confidence levels 2 to 5 according to Schymanski et al. (2014)) (Table S5). Molecular formulas were assigned to ten of these compounds allowing for a 5 ppm mass error. Several polyethylene glycol (PEG) oligomers (octaethylene glycol, decaethylene glycol, dodecaethylene glycol) and 1,6-dioxacyclododecane, 7,12-dione (cycloalkanedione) were semi-quantitatively detected in flowback water with confidence levels 2 to 4 (Table 1).
PEG oligomers, which are common components of fracturing fluids, were previously detected in flowback and produced water from Denver-Julesburg Basin (USA) (Thurman et al., 2017). 1,6-dioxacyclododecane, 7,12-dione is detected in the fracturing fluid for the first time and most probably originates from alkanes naturally present in formation.
Table 1. Individual organic compounds detected in flowback water with the confidence levels 2 to 4 before and after treatment.
2-(2-butoxyethoxy)ethanol, possible transformation product of fracturing fluid additive 2-butoxyethanol, was the only compound identified by suspect screening at the concentration of 29.4 μg/L IS-eq, being an evidence of downhole transformation of this frequently used and persistent fracturing chemical (Table 1).
High concentrations of easily degradable organic compounds, such as acetic acid and ethanol, indicate a potential for biological treatment. Moreover, combination of acetic acid as an energy-limited and ethanol as an energy-excess substrate should lead to higher bacterial growth rates and lower residual concentrations of organic matter (Babel, 2009).
DAF pre-treatment of flowback water
Pre-treatment of flowback water with DAF assisted by coagulation did not change COD and DOC significantly (4.2% and 0.4% removal respectively). However, removal of biopolymer fraction of DOC (74.1%) was observed, probably due to coagulation (Fig. 2). At the same time, increase of the HOC fraction was observed, probably because of the formation of metal-organic complexes which increase hydrophobicity of some organic compounds in flowback water.
Fig. 2. Removal of different DOC fractions after DAF, ozonation, application of GAC (A) and aerobic biodegradation (B).
The pH increase from 4.9 to 8.0 caused by pH adjustment with NaOH required for coagulation has important implications for the following treatment steps. Neutral pH promotes biodegradation of organic matter compared to acidic pH of untreated flowback. The pH of the solution also governs the choice of activated carbon surface chemistry, since it may improve or impair removal of charged organic molecules (Al-Degs et al., 2008).
Application of ozone for flowback water treatment
The studied flowback water had low ozone demand with 44 mg O3/L consumed during 60 min, whereas 300 mg O3/L was applied. Foaming was observed at the beginning of the experiment, possibly caused by the presence of surfactants originated from fracturing fluid. COD and DOC removal was below 10%, as well as removal of different DOC fractions (Fig. 2; Table S4).
Similar results were achieved by Lester et al. (2015) in the experiments with different advanced oxidation processes (UV/H2O2, O2/H2O2, solar light/chlorine, photo-Fenton) treating shale gas flowback water. The authors have shown that OH˙ radicals generated during ozonation were scavenged by bromide, present in the flowback water at high concentration. However, despite high bromide concentrations (570 mg/L) and pH neutralization during DAF pre-treatment, no changes in bromide concentrations were observed even at the highest ozone dosage applied in this study. Low efficiency of ozonation can be thus explained by saturated nature of organic compounds (von Gunten, 2003).