Characterization of flowback and produced water from shale oil and gas wells fractured with different levels of recycled water

Oil and gas production in the United States has increased over the last two decades (EIA, site) owing to increased drilling and production efficiency and the development of new wells and oil and gas plays (Lieskovsky & Gorgen 2013; Kim et al. 2023). The rising domestic production of oil and gas will assist in meeting the expected annual growth of energy and electricity demands of the United States, which are estimated at 0.3 and 0.8% until 2040 (EIA 2015). Furthermore, technically recoverable unconventional oil and gas, amounting to 223 billion barrels and 2,431 trillion cubic feet (Fitzgerald 2013; Gale et al. 2022), respectively, will help provide domestic energy to meet growing demands; moreover, the United States is expected to export as much energy as it exports by the year 2028 (EIA 2015).

Natural gas and oil were extracted from unconventional and conventional reservoirs. Gas and oil in unconventional reservoirs are hard to flow due to the low permeability. Recently, hydraulic fracturing technology has developed rapidly. Advanced hydraulic fracturing has led to the development of extraction gas and oil in unconventional production. Hydraulic fracturing mainly involves the use of water and sand to create fractures and a proppant to maintain the space created in shale or other formations. However, trapped shale oil and gas are released too slowly for large-scale economic applicability (Lieskovsky & Gorgen 2013; Mojid et al. 2021). Horizontal drilling is primarily used for hydraulic fracturing owing to the economic benefits of reduced drilling costs (Lieskovsky & Gorgen 2013; Yang et al. 2022), but this requires high pressure and large amounts of water, which implies that production companies need to have access to ample water sources (Gregory et al. 2011). Specifically, approximately two to seven million gallons of water are required for each well (Lee et al. 2011; Ranm 2011; Stephenson et al. 2011; Nicot & Scanlon 2012; Suarez 2012; Goodwin et al. 2013; Hickenbottom et al. 2013). This water is mixed with chemical additives and sand to ensure that fractures in the formation remain open when injected under high pressure (USDOE 2009; Lee 2011; Spellman 2012).

Pressure arising from local water resource scarcities and public opinion is prompting producers to evaluate their water use plans (FracFocus 2015; Freyman 2014). Nationally, 47% of oil and gas wells that use hydraulic fracturing are located in areas under high (or intensely high) water stress. For example, all wells utilizing hydraulic fracturing in the Denver–Julesburg basin (the area with the highest oil and gas exploration density in the Colorado front range) are classified as being within an area of extremely high water stress (EIA 2014). It is possible to reduce water usage by reusing the flowback and water produced from wells for hydraulic fracturing at other wells, and most oil and gas companies have started reusing flowback water by mixing it with fresh water for hydraulic fracturing (USDOE 2009a; USDOE 2009b; USDOE 2010; Spellman 2012; Jenner & Lamadrid 2013; Lutz et al. 2013; Knapik et al. 2021). However, flowback and produced water have water quality issues, such as high salinity and hardness, and it is difficult to reuse them while maintaining the necessary quality of the fracturing fluid (Lebas 2013; OAQPS 2014; Sareen et al. 2014; Zhou et al. 2014). These water quality issues have created problems with preserving the efficiency of emulsion friction reducers (OAQPS 2014). In addition, treating produced water for its potential reuse presents further problems given its many variabilities that result from factors such as production stimulation methods, geographical location, geological formation, chemicals used during drilling and stimulation, formation depth, hydrocarbon type, and the maturity of the field (Rimassa et al. 2009). Nevertheless, despite these difficulties, utilizing produced water in fracturing operations can improve production from oil and gas wells (Zhou et al. 2014).

Carter et al. (2013) used liquid chromatography-mass spectroscopy (LC–MS) to examine the produced water of the inflow and effluent from a water treatment facility, and the mass spectra of the two samples were almost identical. Kim et al. (2020) used LC–MS to investigate the characteristics of organic compounds in flowback and produced water when recycled and fresh water with fracturing additives were used. However, to the best of our knowledge, the quality of flowback and produced water reused with different ratios of fresh water for hydraulic fracturing has not been investigated to date. Therefore, this study aimed to (1) determine and analyze the water quality parameters of flowback and produced water mixed with varying levels of recycled water over time and (2) use LC–MS to characterize the organic constituents in the flowback and produced water containing different freshwater ratios for various well ages.

The three wells were within 100 m of each other, and the characteristics of their formation were roughly identical. However, each well used varying levels of fresh water to stimulate the formation, as detailed in Tables 1 and 2. Freshwater and recycled water are mixed at the beginning of the hydraulic fracturing process.

FracFocus.org is a registry for the fracturing additives used in hydraulic fracturing nationwide, and it was used here to analyze the level of total dissolved solids (TDS) in the influent, considering that recycled water likely had the largest effect on each well. The detailed compositions of the fracturing fluid of each well are presented in Supplementary material, Table S1.

Well F was fractured using only fresh water and represented the baseline conditions in this study; wells R1 and R2 were fractured with different ratios of recycled water in the fracturing fluid: 1:7 (∼15%) recycled water in R1, and 1:5 (20%) recycled water in R2. These wells were analyzed to determine the impact of recycled water and its higher TDS level on the production of oil and gas and the quality and quantity of flowback and produced water. Other than the ratio of recycled water, the fracturing fluids contained identical ingredients, including an activator, breaker, buffer, crosslinker, proppant, friction reducer, gelling agent, and surfactant.

Following the initiation of flowback, 19 flowback and produced water samples were collected from the three wells across 36 days. The first sample was obtained from the wellhead, and the others from an oil–water separator. The samples were collected daily for the first 14 days, then every 3 days from day 14 to day 29; the final sample was collected on day 36. Samples were immediately transferred to the laboratory and stored at 4 °C before testing. For each collection, 1 L was collected in polyethylene bottles for testing at Colorado State University; 250-mL polyethylene bottles were used for analytical testing at an accredited outside laboratory, and glass vials filled to the brim were collected for volatile compounds analysis.

Conductivity and pH were immediately analyzed onsite after collecting the samples using Hach probes CDC401 and PHC10105, respectively. Standard methods 2320B and 2540 were used to analyze total suspended solids (TSS), TDS, and alkalinity (EIA, 2015). A Shimadzu TOC-VCSH analyzer was used to measure total TOC and dissolved organic carbon (DOC), and a Whatman filter (1.5-μm-equivalent pore size) was used to measure TDS, TSS, and DOC. In accordance with EPA Method 180.1, a Hach 2100N turbidimeter was employed to analyze turbidity, and a Hach DR/4000 spectrophotometer was used to analyze ultraviolet (UV) absorbance at 254 nm (UV254). Prior to determining the concentrations of aluminum, barium, boron, calcium, iron, potassium, magnesium, sodium, silica, strontium, and zirconium using inductively coupled plasma atomic emission spectroscopy, the pH level was adjusted to <2. Chloride concentrations were obtained via silver nitrate titration in accordance with EPA Method 9253, and bromide concentrations were obtained using an ion chromatograph in accordance with EPA Method 300. Ammonia and sulfate concentrations were obtained following EPA Method 350.1 and ASTM Method D516, respectively.

In this linear regression model, the alternative hypothesis was that the coefficient β_i was not zero and the null hypothesis was that the coefficient β_i was zero. Coefficients β₁ and β₂ were fitting constants for Well F, coefficients β₃ and β₄ were fitting constants for Well R1, and coefficients β₅ and β₆ were fitting constants for Well R2.

Flowback water samples from the three wells were collected on days 1, 2, 6, 10, 14, and 20 for LC–MS using an Agilent 6530 quadrupole time-of-flight with electrospray ionization in a positive setting. Other settings were as follows: sheath gas flow, 12 L/min; shear gas temperature, 400 °C; gas flow, 12 L/min; nebulizer pressure, 30 psig; gas temperature, 325 °C; octupole RF peak voltage, 750 V; fragmentor voltage, 120 V; skimmer voltage, 60 V; nozzle voltage, 500 V. The carrier flow rate was set to 0.3 mL/min through the C-18 column (2.1 × 100 × 2.7 mm), and 0.1% formic acid in water and 0.1% formic acid in acetonitrile were applied to mobile phases A and B, respectively. A gradient of 95–80% A was applied for 1–8 min, one of 80–5% A was applied for 8–17 min, and one of 5–95% A was applied for 17–18 min. A total of 5 μL was injected.

There were no significant variations between the organic constituent contents (including gasoline range organics (GRO), diesel range organics (DRO), oil range organics (ORO) and total petroleum hydrocarbons (TPH)) of the wells (Supplementary material, Figure S1). All organic constituents showed significantly different temporal variations, which were more pronounced at the beginning of the sampling period. TOC showed much greater variations than alkalinity (Figure 2(b)). The TOC level and organic constituent levels fluctuated significantly throughout the sampling period, but there were no significant differences among wells, suggesting that these parameters were unaffected by the recycled water ratio. The average TOC levels of Marcellus wells, which are usually fractured with slickwater fluids, are typically <250 mg/L, and show decreasing trends or stabilization with time (Maguire-Boyle & Barron 2014; Kim et al. 2016). Furthermore, while TOC levels differ significantly among wells that use different types of hydraulic fracturing fluid, TOC levels still decrease or stabilize over time (Kim et al. 2019). These reported trends agree with the TOC levels in this study.

The pH data points were highly sporadic, and this inconsistency continued throughout the sampling period. However, the overall trend in the data showed a decline toward neutral pH over time (Figure 2(c)). Well F became stable after day 13, varying between 7.7 and 7.2; Well R1 was the most stable over the first 20 days but fluctuated over the last 10 days; and Well R2 varied throughout the sampling period. Interestingly, pH was not affected by the base fluid and recycled water ratio. One possible explanation for the similar pH levels in the wells over time may be geological formation buffering and flowback water ionic composition. Previous work has reported that significantly different initial pH values in fracturing fluid do not influence the pH of flowback and produced water (Kim et al. 2016).

The differences between the produced waters of the three wells (the most notable difference was noted in TDS; Figure 2(d)) were likely due to variations in the recycled produced water used in the fracturing fluids. TDS differed significantly among the wells (p-value < 0.05) and was observed to increase over time. The significant difference between the TDS levels of the wells and their observed temporal increments may have been due to the different recycled water ratios and the increasing impact on water quality parameters associated with the geological formation. The observed increasing trend in TDS agreed with that observed in previous studies that highlighted significant variations between sampling fields and reported different magnitudes of stabilized TDS levels (Kim et al. 2016; Kim et al. 2019).

The concentrations of several inorganic elements (including boron, barium, bromide, calcium, iron, magnesium, and strontium) differed among the wells (p-values < 0.05; Supplementary material, Figures S2 and S3). The differences may not be related to the fracturing fluid itself but rather to chemical interactions with the formation (Barbot et al. 2013; Kim et al. 2016) or from the different initial fracturing fluids containing various inorganic concentrations stemming from different recycled water ratios. Aluminum and sulfur were the only inorganic elements for which the differences between wells were not statistically significant.

Aluminum, sulfate, and zinc did not vary significantly over time (Supplementary material, Figures S2 and S3). The potassium concentration in the produced water differed significantly among the three wells but did not vary over time, suggesting that the potassium concentrations in the produced water from each well temporally increased and declined at identical rates but had significantly different values. The zinc concentrations increased and declined consistently, while those of the other elements significantly differed temporally, with p-values < 0.05. Ammonium and silicon varied significantly over time.

Estimating the coefficients for flowback and produced water quality is difficult because many factors (such as fracturing additives, temporal and spatial variability, vertical well depth, and the ratio of fresh and recycled water) can affect their quality (Barbot et al. 2013; Kim et al. 2016). Since the same organic compounds were used as a fracturing additive for drilling the three wells in relatively close proximity, the spatial variability and fracturing additives were assumed to be insignificant. Therefore, it was assumed that the temporal variation and ratio of fresh and recycled water affected all the major water quality parameters.

The linear regression analysis results (Table 3) showed that there were statistically significant differences between the coefficients β₃ and β₄ of wells R1 and F, while coefficients β₅ and β₆ indicated a statistically significant difference between wells R2 and F. The correlation of inorganic elements with time was also studied. Comparably lower values of the coefficient of determination (R²) indicated that no higher correlation existed between the water quality parameters and temporal variability. However, the analysis indicated relatively positive correlations between key ion concentrations (bicarbonate, calcium, magnesium, and strontium) in flowback and produced water with time (with R² values > 0.70), but this was not true for chloride and sodium. The highest correlation was obtained for strontium. These results showed that the concentrations of key ions increased with time, meaning that their source was the formation itself (Kim et al. 2016). No correlations between silicon and potassium and temporal variability were observed, with very low R² values, even though they are usually used as fracturing additives in hydraulic fracturing (Kim et al. 2016; Kim et al. 2019). For the initial flowback period of four weeks, almost 50% of the flowback volume was recovered and 50% of the fracturing additives remained in the formation (Kim et al. 2016). The lack of a correlation between the main fracturing fluid ingredients and temporal variables may have been due to most of the fracturing additives remaining in the formation during the early flowback period.

The p-values of barium and bromide for coefficients β₃, β₄, β₅, and β₆ were >0.05, as shown in Table 3, indicating that the fitting constants were not significant. Therefore, coefficients β₃, β₄, β₅, and β₆ were set to zero, inferring that there were no statistically significant differences in the barium and bromide concentrations in the flowback and produced water from the three wells. Similarly, the p-values of bicarbonate, calcium, and magnesium for coefficients β₃, β₄, and β₆ were >0.05, meaning that bicarbonate, calcium, and magnesium also showed no significant differences among the wells, except for well R2, whose intercept was coefficient β₅.

The p-values of chloride for coefficients β₄ and β₆ were zero, indicating that the intercepts of the wells differed while their slopes were identical. These results indicated that the different intercepts may have resulted from differing initial chloride concentrations associated with the freshwater ratio. The identical slopes for chloride inferred that the chloride concentration increased at the same rate in all wells, and this was likely because it was affected by the geological formation.

The value of sodium for coefficients β₄, β₅, and β₆ approached zero, meaning that wells F and R2 were statistically identical except for the different intercepts. The p-values of calcium, chloride, and sodium for β₄ and β₆ were relatively high, suggesting similarity among the three wells. However, the p-value of sodium for β₃ was marginally <0.05, indicating that at the beginning of water extraction, the variation in the sodium concentration was relatively high in wells F and R1. Overall, calcium, chloride, and sodium exhibited similar trends in the three wells. These elements were considered to be the dominant inorganic compounds in the flowback and produced water, and they exhibited a slightly increasing trend over time due to the influence of the geological formation (Barbot et al. 2013; Kim et al. 2016). The main inorganic compounds appeared to be more affected by the geological formation than the recycled water ratio. However, only β₁ of the main ingredients of the fracturing additives, including aluminum, boron, and silicon, was significantly constant, indicating that the main ingredients had constant values with similar temporal variabilities. These results indicated that the ions used as the main hydraulic fracturing ingredients were affected by the initial concentration of the fracturing fluid rather than by the recycled water ratio. Recycling of flowback and produced water is being sought by most oil and gas companies from many parts of the country. This trend anticipates reducing concerns of regional water scarcity due to hydraulic fracturing. An in-depth understanding of produced water quality when recycling produced water with fresh water with varying ratios, can improve wastewater treatment process strategies, and minimize concerns of water scarcity.

Agilent Technology Software was used to conduct further analysis and identify the organic compounds in the three wells. Owing to the limited available chemical standards for the qualitative analysis of organic compounds used for hydraulic fracturing, a library searching method in Agilent Technology Software was used, and 1 ppm mass accuracy was applied to match chemical compounds. Varying with time, 3,4,4-trimethyloxazolidine, acrylic acid, 2-hydroxyethyl ester, aldol, benzenecarboperoxoic acid, 1,1-dimethylethyl ester, butyl glycidyl ether, decyl-dimethyl amine oxide, di (ethylene glycol) ethyl ether acetate, dibutylaminoethanol (2-dibutylaminoethanol), diethylbenzene, dipropylene glycol, dipropylene glycol monomethyl ether (2-methoxymethylethoxy propanol), d-Limonene, ethanol, 2, 2′-(Octadecylamino) bis-, ethoxylated oleyl amine, ethyl acetoacetate, furfuryl alcohol, glutaraldehyde, phthalic anhydride, polyethylene glycol, salbuterol, triethylene glycol, and triisopropanolamine were observed in samples from the three wells using an LC–MS–ESI-positive mode.

Reusing produced water for hydraulic fracturing additional wells is considered as a good approach to handling the wastewater. Conventional wastewater treatment processes are usually designed for removing particles and hence high levels of organic remained. In-depth knowledge of organic compounds in produced water when blended with different ratios of recycled water and fresh water, can improve water treatment process strategies for managing organic composition in produced water from hydraulic fracturing.

We analyzed the organic and inorganic compounds in flowback and produced water samples from three wells fractured with different fresh-to-recycled water ratios. The TDS in all three wells gradually increased over a 36-day period, but the amounts differed significantly between the wells. There were no significant differences between the TOC and organic constituents, including GRO, DRO, ORO, and TPH, between wells, and a slightly decreasing trend was observed over time. LC–MS with electrospray ionization was used to compare and identify the organic compounds in the flowback and produced water samples of the wells over time. Ethoxylated functional units were detected in all three wells, presenting similar masses but different spectral patterns. A qualitative analysis was performed using the exact mass of the chemical compounds generally used in hydraulic fracturing processes, and similar organic compounds were observed in the flowback and produced water samples of the three wells. Notably, this work focused on a restricted set of wells with a small number of samples. Future research should be conducted to focus on a larger set of wells. The results of this study provide new insights into the quality of flowback and produced water from wells fractured using different fresh-to-recycled water ratios and can be used as a reference for improving the reusability of treated water and reducing wastewater discharge.

This research was supported by the National Research Foundation of Korea (NRF) [grant number 2021R1C1C101040411]. This research was supported by “Regional Innovation Strategy (RIS)” through the National Research Foundation of Korea(NRF) funded by the Ministry of Education(MOE)(2021RIS-002).

A.F. and C.F. conceptualized the study, and wrote, reviewed, and edited the article. E.F. wrote, reviewed, and edited the article.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.