Article pubs.acs.org/est
Assessment of Effluent Contaminants from Three Facilities Discharging Marcellus Shale Wastewater to Surface Waters in Pennsylvania Kyle J. Ferrar,†,‡,* Drew R. Michanowicz,†,‡ Charles L. Christen,†,§ Ned Mulcahy,† Samantha L. Malone,†,‡ and Ravi K. Sharma†,⊥,∥ †
Center for Healthy Environments and Communities (CHEC), Graduate School of Public Health (GSPH), University Of Pittsburgh, 4200 Fifth Avenue Pittsburgh, PA 15260 ‡ Department of Environmental and Occupational Health (EOH), GSPH, University of Pittsburgh, 4200 Fifth Avenue Pittsburgh, PA 15260 § Pittsburgh AIDs Taskforce; ∥Department of Behavioral and Community Health Science (BCHS), GSPH, University of Pittsburgh, 4200 Fifth Avenue Pittsburgh, PA 15260 ⊥ Department of Biostatistics, GSPH, University of Pittsburgh, 4200 Fifth Avenue Pittsburgh, PA 15260 S Supporting Information *
ABSTRACT: Unconventional natural gas development in Pennsylania has created a new wastewater stream. In an effort to stop the discharge of Marcellus Shale unconventional natural gas development wastewaters into surface waters, on May 19, 2011 the Pennsylvania Department of Environmental Protection (PADEP) requested drilling companies stop disposing their wastewater through wastewater treatment plants (WWTPs). This research includes a chemical analysis of effluents discharged from three WWTPs before and after the aforementioned request. The WWTPs sampled included two municipal, publicly owned treatment works and a commercially operated industrial wastewater treatment plant. Analyte concentrations were quanitified and then compared to water quality criteria, including U.S. Environmental Protection Agency MCLs and “human health criteria.” Certain analytes including barium, strontium, bromides, chlorides, total dissolved solids, and benzene were measured in the effluent at concentrations above criteria. Analyte concentrations measured in effluent samples before and after the PADEP’s request were compared for each facility. Analyte concentrations in the effluents decreased in the majority of samples after the PADEP’s request (p < .05). This research provides preliminary evidence that these and similar WWTPs may not be able to provide sufficient treatment for this wastewater stream, and more thorough monitoring is recommended.
■
hydraulically fractured. Hydraulic fracturing fluid is comprised of large volumes of water18 containing about 1% chemical additives19 and proppant (typically silica), which is injected into a well’s borehole at high pressures.20 The injections fracture the shale enabling the propagation of deep shale fissures, resulting in a pressurized environment conducive to the liberation of gas and materials trapped in the shale.8,21−25 Anywhere from 10% to 80% of the injected volume may return to the surface as wastewater. The wastewater from this entire process includes both “flowback” and “produced water.” Flowback is the fracturing fluid that quickly returns to the surface, and produced water is the
INTRODUCTION
Unconventional natural gas development (UNGD), which made up 23% of natural gas production in 2010 in the United States, is increasing each year and is expected to grow to 49% by 2035.1 A number of concerns arise from the hydraulic fracturing process including explosion risks,2 air,3−7 water8−10 and soil pollution, and a range of social impacts.11−14 This paper focuses on the treatment of UNGD wastewaters by wastewater treatment plants (WWTPs), and the subsequent discharges to surface waters. The wastewater stream is a separate topic with its own research needs,15,16 and research shows the potential risk of environmental contamination during disposal of UNGD wastewaters is several orders of magnitude larger than other pathways of exposure.17 Gas extraction from tight shales such as the Marcellus Shale is only economically viable after natural gas wells are © XXXX American Chemical Society
Received: April 10, 2012 Revised: February 19, 2013 Accepted: March 4, 2013
A
dx.doi.org/10.1021/es301411q | Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology
Article
fracturing fluid that takes longer to return to the surface.26,27 Both are enriched with materials from the shale formation, such as brines, hydrocarbons, and naturally occurring radioactive materials, and the longer the fluid takes to return to the surface the greater the concentration of formation materials.15,28−31 Initially, UNGD wastewaters in the Marcellus Shale region was processed predominately by municipal sewage treatment plants (publicly owned treatment works), even though the wastewater has been documented to disrupt microbial digestion processes at high concentrations.32 Another predominant treatment option has been industrial wastewater treatment plants designed to treat said wastewater. Both methods discharge treated effluent into surface waters.33 The surface water discharges from these WWTPs have been linked to increasing concentrations of bromides and total dissolved solids (TDS) in surface waters of Pennsylvania.34,35 In response, public water systems (PWSs) have switched from chlorine to chloramine for disinfection.36 Chloramine reduces the formation of certain disinfectant byproducts (DBPs), but may increase the formation of other unregulated/ undetected DBPs and can increase lead (Pb) exposure.10,37,38 In May 2011 the Pennsylvania Department of Environmental Protection (PADEP) requested that UNGD drillers cease discharging the wastewaters into surface waters through WWTPs, although compliance with the request was voluntary.39 Then in March 2012, updated permitting standards were introduced for WWTPs discharging the wastewater to surface waters.40 Treatment and reuse (recycling) is the industry’s current best practice, but these methods are still being developed and have not been comprehensively adopted in the field.15,41 While underground injection into Class II injection wells has increased in other states, there are few injection wells in Pennsylvania.41 As a result, more wastewater has been shipped out of Pennsylvania, specifically to Ohio.33,42 Research has not been conducted to evaluate the environmental health impact of UNGD wastewaters discharged from WWTPs.42 Therefore, this research focuses on the preliminary step of characterizing the WWTP effluent containing UNGD wastewaters. Effluent samples were collected from the outfalls of three WWTPs while wastewater was being treated and discharged. Our first hypothesis was that analytes were present at concentrations above water quality criteria concentrations directly in the effluent. Follow-up sampling was conducted at each facility after the PADEP’s request. Our second hypothesis was that analyte concentrations would decrease, following compliance with the PADEP’s request.
scintillation vials, with care to eliminate head space. All vessels were rinsed three times with effluent before the sample was collected. WWTP-1 was sampled on three separate occasions, twice from the effluent, and once from the influent. The first effluent sampling began at 19:00 on October 19, 2010 (sample set 1A, n = 24), when the UNGD wastewater was believed to enter the primary treatment with municipal influent. Samples were collected hourly, for a full 24 h period. On November 10, 2010, a single influent sample was taken from WWTP-1’s UNGD wastewater holding cell at 13:00 (Sample set 1-Influent, n = 1) (Table 2). On December 1, 2011, 9 samples were collected, over a 24 h period beginning at 17:00, with 1 sample collected every 3 h (sample set 1B, n = 9). At WWTP-2, sampling was also conducted for a full 24 h period, beginning at 15:00 November 10, 2010 (sample set 2A, n = 24). Nine more samples were then collected over a 24 h period beginning November 7, 2011 at 17:00, with 1 sample taken every 3 h (sample set 2B, n = 9). At WWTP-3, sampling was conducted on three separate dates. The first sample set was collected December 10, 2010 beginning 11:00 (sample set 3A, n = 8) and the second beginning 13:00 April 1, 2011 (sample set 3B, n = 8). Both included eight samples, 1 collected every 3 h for a full 24 h. WWTP-3 was sampled again May 25, 2011. Rather than a 24 h period, eight samples were collected over 8 h, 1 sample taken every hour beginning 11:00 (sample set 3C, n = 8). Analyte selection focused on constituents characteristic of UNGD flowback and produced waters, as well as analytes of specific public health concern. Potential analytes were identified by reviewing a Marcellus Shale Coalition report on produced water and the U.S. Geological Survey database for produced waters.29,43 The list was narrowed down to include the major ions in the wastewater, elements that pose a potential threat to public or environmental health at elevated levels, and several organics commonly found in produced water. The full list of analytesparticularly organicsanalyzed by Hayes (2009), is significantly large and was beyond the scope of this preliminary study.29 Organic analytes were not detected during initial sampling at WWTP-1 and WWTP-2 (sample sets 1A and 2A, respectively), therefore they were not included in the suite of analytes during the follow-up sampling sets (1B and 2B). Analytes were measured for total metal content to compare to U.S. Environmental Protection Agency (USEPA) MCLs and human health criteria (for the consumption of “organism” or “water and organism”), which are based on “total concentration” sampling data. If metal analytes were to be compared to aquatic life standards, samples would have required filtering with a 0.45 μm filter following collection to obtain “dissolved concentration.” Aquatic life concentration standards do not exist for the metals of concern, therefore sample filtration was not necessary. Turbidity measurements for sample sets 1A, 2A, and 3A were taken during sample collection using a Hanna Instruments 93707 turbidity meter. The samples were analyzed by the RJ Lee Group, Murrysville, PA according to the following USEPA approved methods: 200.7 for inorganics, barium (Ba), calcium (Ca), magnesium (Mg), manganese (Mn); 200.8 for strontium (Sr); 300.0 for inorganic anions bromide (Br), chloride (Cl), and sulfate; 624 for organics benzene, ethyl benzene, toluene, xylene; and 8015 for 2-butoxyethanol (2-BE). TDS was analyzed using method SM2540C. The statistical analysis was conducted using SPSS Statistics Version 19 (Armonk, NY: IBM Corp). Concentrations data were explored for aberrations using descriptive statistics with
■
MATERIALS AND METHODS Sampling sites included the City of McKeesport Municipal Authority (WWTP-1), the Franklin Twp. of Greene County Municipal Authority (WWTP-2), and the Hart Resources PA Brine Josephine treatment facility (WWTP-3). An influent sample from WWTP-1 was also collected. For site characterization, each WWTP was toured with facility managers who provided detailed explanations of the treatment processes. For additional background information, file reviews were conducted at the southwestern PADEP offices. Facility permits, discharge monitoring reports (DMRs), and facility correspondences were reviewed. Facility descriptions are presented in the Results. Repeated grab samples (dt = 1−3h) were taken directly from the effluent stream at the end of each WWTPs’ actively discharging outfall pipe, before it mixed with the surface water. Inorganic samples were collected in 1 L polyethylene sample bottles with screw tops. Organic samples were collected in glass B
dx.doi.org/10.1021/es301411q | Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Environmental Science & Technology
Article
0.14% of the effluent for the total daily volume. Since the total volume of wastewater was released at once rather than throughout the whole day, an estimation of the actual dilution may be 0.81% UNGD wastewater in the effluent as it was discharged 8−12 h later. WWTP-1 was not treating UNGD wastewater when the site was sampled on December 2, 2011 (1B). The Franklin Twp. of Greene County publicly owned treatment works (WWTP-2) accepted UNGD wastewater through primary treatment with the municipal influent. The wastewaters were piped into the sewer system after undergoing a pretreatment process at the Tri-County Wastewater facility. Pretreatment focused on removing total suspended solids (TSS) and “oil and grease” from the wastewater by filtration, flocculation, and skimming. These treatments are expected to reduce organic loads, but not impact TDS or dissolved inorganic elements. Treatment at WWTP-2 included settling, filtration, anaerobic and aerobic digestion, and clarification before discharging into the lower fork of Ten Mile Creek, a tributary of the Monongahela River in Waynesburg, PA. WWTP-2 treated the wastewater continuously and was allowed to accept 50 000 gallons per day (GPD) of UNGD wastewater from Tri-County Wastewater. According to the facility’s files, the Tri-County pretreatment plant closed on March, 21 2011, thus WWTP-2 no longer accepts UNGD wastewater. During the first sample set (2A) at WWTP-2, the facility treated 53,200 gallons of UNGD wastewater November 10, 2010 and 52 600 gallons November 11, 2010. The average effluent flow was 0.982 million gallons per day (MGD) for this period, meaning UNGD wastewater comprised approximately 5.4% of the effluent by volume. Over the second sampling period, November 7, 2011 (2B), the facility was not treating UNGD wastewater. The PA Brine Josephine facility (WWTP-3) exclusively treats oil and gas wastewater, both conventional and unconventional. Wastewater is hauled on-site via 5000-gallon residual waste tanker trucks and is stored in enclosed storage tanks. The large intake results in continuous wastewater influent treatment and effluent discharge. Debris is initially removed in an open spillway, and then treatment begins via settling in an enclosed vessel. Sodium sulfate (Na2SO4) and a polymer agent are then added to the oil and gas wastewater to aid the precipitation of barium and other alkaline earth metals such as strontium. Fine lamellae screens filter and clarify the wastewater before a final silicone defoamer agent is added prior to surface water discharge. The solid waste is dried with a mechanical press and trucked to residual waste landfills. These processes are expected to precipitate dissolved cations and filter solid elements, but are not expected to impact dissolved anions, such as chlorides. The filters are expected to decrease TSS, but would not have an impact on TDS. After treatment, the effluent is discharged into Blacklick Creek, a tributary of the Conemaugh River within the Allegheny River Basin. When WWTP-3 was sampled on December 10, 2010 (3A) and April 1, 2011 (3B), the facility manifests showed the majority of the influent oil and gas wastewater was from UNGD. On the third sampling date, May 25, 2011 (3C), the majority of wastewater was from conventional oil and gas operations, according to facility correspondences with the PADEP. As of September 19, 2011, the facility claims to have stopped accepting UNGD wastewaters, although they reported treating 332 166.7 gallons of UNGD wastewater between June and December 2011.49 The volume accepted and treated January-June 2012 has not yet been published.
box plots and histograms. Mean and maximum analyte concentrations from each sample set were compared to water quality criteria (Table 1). The criteria included both enforceable and nonenforceable drinking water standards published by the USEPA: primary drinking water regulations known as maximum contaminant levels (MCLs), secondary drinking water regulations known as secondary maximum contaminant levels (SMCLs),44 surface water concentrations set for human health criteria for the consumption of aquatic organisms, and aquatic life criteria known as criterion maximum concentration (CMC) and criterion continuous concentration (CCC).45 For strontium, a nonenforceable recommended concentration for drinking water of 4 mg/L was used.46 For bromide, a concentration value of 0.1 mg/L that poses a risk for DBP production at PWSs was taken from the literature.47 Analyte sample sets for each facility were split according to sampling date, as ‘before’ or ‘after’ the PADEP’s voluntary compliance request on May 19, 2011. Intrafacility comparisons were conducted using Mann−Whitney tests for each analyte. The Mann−Whitney test was chosen because sample sizes were small, and many of the analyte concentrations were not normally distributed. For the statistical comparisons, nondetected analytes in sample sets were given a value of (1/2) the method detection limit (MDL). To address serial correlation, serial dependence of samples due to the repetitive sampling scheme, the data was “pre-whitened” prior to the Mann− Whitney tests, according to the methodology outlined by Yue and Wang (2002).48 Time series plots were created for each analyte sampled at all three sites (Figures 1, 2, and 3). Each plot shows analyte concentrations, both before and after the WWTPs were asked to stop treating UNGD wastewater. Estimations of influent concentrations were calculated for WWTP-1 and 2 using the reported dilution factors (Table 2) and subtracting the follow-up analyte concentrations (sample sets 1B and 2B) from first sample set concentrations (1A and 2A, respectively). Descriptive statistics are included in the “Supporting Information” section in Tables S1−S8. Complete sampling data is presented in Table S9.
■
RESULTS WWTP-1, the City of McKeesport’s publicly owned treatment works, accepted UNGD wastewaters via 5000 gallon residual waste tanker trucks, according to the facility manager. The PADEP allowed WWTP-1 to accept 1% of its daily flow of UNGD wastewater. The wastewater was pumped from the trucks and kept in a large, open holding tank until released into the primary treatment process along with the municipal influent. The facility manager indicated that the wastewater moved through the treatment process and exited the system 8−12 h later. Treatment processes include physical filtration, flocculation, aerobic digestion, and clarification before discharging into the Monongahela River. These processes could be expected to remove organic compounds through degradation, but would not be expected to impact soluble, inorganic elements. However, solid inorganic elements may be removed during the filtration of organic material. According to WWTP1’s facility manifests, UNGD wastewater was no longer accepted as of April 19, 2011, but the facility continued to accept wastewater from shallow or conventional oil and gas field operations. On the first sample date at WWTP-1, October 19, 2010 (1A), the facility treated 13 020 gallons of UNGD wastewater, with an average daily flow of 9.6 million gallons per day (MGD). The UNGD wastewater comprised approximately C
dx.doi.org/10.1021/es301411q | Environ. Sci. Technol. XXXX, XXX, XXX−XXX
1.63
0.60
228.7
98.1
562.2
strontium (mg/L)
bromide (mg/L)
chloride (mg/L)
sulfate (mg/L)
TDS (mg/L)
11.2
648
139
377
0.944
2.26
0.021
D
