1. Introduction
The fastest growing energy sector in the US is unconventional shale gas and oil production. Hydraulic fracturing combined with horizontal drilling is used for exploitation of tight rock formations containing abundant oil and gas resources that were previously unreachable [
1]. The extraction of shale gas using advanced hydraulic fracturing has increased from 14% of U.S. natural gas production in 2004 to 97% in 2018. However, this increase has led to a concurrent increase in water usage [
2,
3]. In hydraulic fracturing, water, mixed with chemicals, is pumped at high pressure through the well bore to fracture the tight rock formation. Subsequently, the pressure is reduced, and the water flows back to the surface as flowback and produced water, known collectively as PW. Around 15–23 million liters of PW are generated during the extraction period of each well [
4]. Approximately 116 billion liters of PW are produced in the U.S. annually [
5]. About 20.06 million liters of water were used per well in the Fayetteville shale [
6].
Water is a very scarce and valuable natural resource. Promoting circularity in water usage is essential in order to develop sustainable manufacturing processes. Recovery and reuse of PW are essential. The work conducted here directly addresses this major societal issue. PW is highly impaired; thus, treating this water is very challenging. It contains a range of contaminants, as well as high total dissolved solids (TDS) concentration, high total suspended solids (TSS), polar and non-polar organic compounds, and low surface tension dissolved species [
7,
8,
9,
10]. Currently, the PW is frequently deep well injected. However, deep well injection practices are non-sustainable [
11] and have several drawbacks, including the limitation of available deep well injection sites, the cost of transporting PW to the available sites, and the possibility of creating earthquakes. Importantly it does not lead to the recycling and reuse of water.
Limited options exist to treat PW. Some investigators have considered distillation-based technologies, such as multistage flash distillation, mechanical vapor recompression, or integrating evaporation, crystallization, and spray drying to treat PW [
12]. Though successful in treating high TDS brines, these technologies suffer from some drawbacks, such as high cost, large footprint, and the use of chemicals [
13]. Here, membrane technology is considered to treat high TDS brines.
Conventional membrane processes such as reverse osmosis (RO) can be used to treat brines with TDS values below 50,000 mg/L [
14]. However, as the osmotic back pressure increases, the amount of water that can be recovered using RO, especially as the TDS increases over 50,000 mg/L, is limited [
14,
15]. Membrane distillation (MD) is an emerging technology that can be used to treat high TDS PW. In MD, a microporous hydrophobic membrane is used as a barrier between the feed and permeate streams. The feed stream is heated relative to the permeate stream. This imposed temperature gradient leads to a vapor pressure gradient across the membrane. Water vapor passes down the vapor pressure gradient from the feed to the distillate. Importantly non-volatile solutes cannot pass through membrane pores. Unlike reverse osmosis, the depression of the feed vapor pressure with increasing feed TDS is much less than the increase in osmotic back pressure [
16,
17]. Here, direct contact membrane distillation is used where the feed and permeate streams are in direct contact with the two surfaces of the membrane [
18].
However, like all other membrane technologies, MD suffers from fouling of the membrane by rejected species. In the case of hydraulic fracturing PW, dissolved polar and non-polar organic species can easily adsorb onto the hydrophobic membrane surface. In addition, low surface energy compounds, such as surfactants, can adsorb onto the membrane. Scaling by dissolved salts at high TDS can also occur. Fouling leads to a drop in permeate flux. However, it can also lead to failure of the membrane whereby water and dissolved non-volatile solutes pass directly through the membrane pores [
19].
Our aim is to develop an integrated process to maximize water recovery and suppress membrane fouling. Commercially available polyvinylidene fluoride (PVDF) membranes have been used to evaluate the feasibility of the integrated electrocoagulation ultrafiltration membrane distillation crystallization (EC/UF MDC) process to address both scaling and wetting of the membrane and to maximize water recovery. Electrocoagulation (EC) is used to pretreat the feed and remove dissolved organic compounds that could foul the membrane. UF is used to rapidly separate the permeate from the EC sludge. MD is used to recover treated water. By linking this with crystallization, we maximize water recovery while suppressing scaling on the MD membrane.
EC is an electrolysis process where a sacrificial electrode (anode) is used to generate metal ions. These metal ions generate a variety of metal hydroxides as follows: M
(s) → M
n+(aq) + ne
− at the anode. Water is reduced at the cathode by the reaction 2H
2O + 2e
− → 2OH
− + H
2, where M is often Al or Fe [
20]. Metal complexes such as M(OH)
(n−1)+, M(OH)
2(n−2)+ and M
6(OH)
15 (6n−15)+ are produced. These metal complexes contribute to the neutralization of the negatively charged organic species and suspended solids. As the solution ages, they convert to amorphous M(OH)
n(s) particles. M(OH)
n(s) particles can easily adsorb and trap organic compounds and suspended solids. EC can effectively remove a wide range of contaminants, including suspended solids, emulsified oils, heavy metals, and some organic compounds. Using EC as a pretreatment step can provide several benefits, including the high removal of contaminants, operational flexibility, production of 80% fewer solids, no hazardous waste disposal cost, and reduced use of expensive chemical agents.
Crystallization after EC UF MD can minimize membrane fouling and scaling by reducing the formation of crystal nuclei in the bulk feed. This is particularly important when using hypersaline PW with a high TDS concentration. In addition, the EC UF MDC process can also offer a potential solution to high TDS brine disposal by recovering both water and minerals, which can lead to a nearly zero liquid discharge [
21,
22]. Here, the feasibility of using EC UF MDC to recover water and minerals from hypersaline shale gas PW has been investigated. In addition, we compared the increase in water recovery when adding a crystallization unit by comparing EC UF MDC with EC UF MD. As about 80% of the water utilized in hydraulic fracturing is surface water and groundwater, the process developed here could have an impact on hydraulic fracturing operations [
23]. Further, about 95% of the collected PW is directly disposed of in a Class II disposal well [
24].
Figure 1 shows the concept of the combined processes. The feed consisted of 3 L. After UF, about 1.5 L of permeate is recovered.
Table 1 shows previous studies on treating synthetic and actual wastewater, including PW, seawater, and RO brines. Previous studies have been conducted using EC MD or MDC systems to treat low TDS brines as well as synthetic PW. To our knowledge, this is the first study that has considered an integrated EC UF MDC system to treat hypersaline PW from unconventional oil and gas wells.
2. Methods
2.1. PW Characterization
PW samples were collected from a hydraulic fracturing facility in Midland, Texas, USA. The samples were analyzed at the Arkansas Water Resources Center, University of Arkansas (Fayetteville, AR, USA). The received PW had been treated with chlorine dioxide at the hydraulic fracturing facility to remove bacteria and iron. EPA standard methods 160.1, 160.2, 415.1, and 180.1 were used to measure TDS, TSS, turbidity, and total organic carbon (TOC) [
31]. EPA methods 200.7 and 300.0 were used to measure cations and anions, respectively. Conductivity was measured using a conductivity meter (VWR, Radnor, PA, USA).
2.2. Membrane Characterization
Static water contact angles were measured using a sessile drop contact angle goniometer (Model 100, Rame-Hart Instrument Company, Netcong, NJ, USA). The DI water droplet volume was 3 µL introduced at a rate of 0.5 µL/s. Each droplet was allowed to stabilize for 10 sec prior to measurement. For each membrane, the average value of three measurements obtained at three different locations was used in this study.
For each membrane before and after MD or MDC, both the surface morphology and elemental analysis were obtained using scanning electron microscopy (SEM) and energy-dispersive X-ray (EDX) spectroscopy, respectively, using Nova Nanolab 200 Duo-Beam Workstation (FEI, Hillsboro, OR, USA).
2.3. EC UF Pretreatment
Figure 2 shows the EC UF system. Here, a custom-built polycarbonate reactor with a volume of 1078 cm
3 (dimensions of 7 cm × 11 cm × 14 cm) was used to conduct all EC experiments. Five aluminum electrodes were fitted vertically inside the reactor. The inter-electrode spacing was 10 mm. The residence time in the reactor was 5 min. A DC power supply (Hewlett Packard, Palo Alto, CA, USA) was connected to a reverse polarity switch which enabled the direction of the current to alternate every 30 s. This step is essential, as shown by our earlier work, to prevent the formation of a passivation layer on the electrode surface which would suppress further reactions [
32,
33].
The first and last electrodes were connected to the power supply in a bipolar series (BPS) configuration to simplify the electrical connections. In previous studies, the BPS configuration was shown to lead to an enhancement in TOC reduction [
34]. In earlier studies, several EC experiments were conducted to determine the appropriate current and reaction time [
34,
35]. Based on these earlier studies, a range of currents from 1 to 3 A and a reaction time of 5 min were studied here.
EC is an electrolysis process where aluminum ions are continuously generated at the anode while the reduction of water takes place at the cathode leading to the formation of hydrogen gas and hydroxide ions [
34,
35]. However, the actual reactions that occur depend on the reduction potentials of the other species present in the feed. A range of poly aluminum hydroxides is produced in the solution when coagulating ions (aluminum and/or hydroxide ions) undergo hydrolysis in water. These aluminum hydroxides can help destabilize suspended, emulsified, and dissolved contaminants which can aggregate and precipitate as sludge or lift up to the surface as flocs.
Firstly, three standalone EC experiments were conducted to optimize the current based on achieving a high TOC reduction. After each EC experiment, treated water was removed from the sludge and settled floc after sedimentation for 30 min. After optimizing the current, the EC reactor was run to treat 3 L of PW. This process takes about 15 min, which leads to an average floc aging time of 7 min. Next, the supernatant from the EC becomes the feed to the UF process.
The UF process was conducted after EC using a ceramic membrane from CeraMem (Waltham. MA, USA). Crossflow filtration was conducted using a 10 nm nominal pore-size ceramic membrane module. The active surface area was 0.13 m2. After EC, the 3 L of PW was placed in the UF feed tank. The feed was recirculated through the module using a diaphragm pump (P800, King-Kong, Triwin, Taichung City, Taiwan), keeping the permeate outlet closed. The permeate outlet was opened after 5 min. The transmembrane pressure (TMP) was 65 kPa at a feed flow rate of 2.5 L/min. The permeate water was collected in the permeate tank, which was placed on a balance (Mettler Toledo, Columbus, OH, USA). The permeate flux was calculated based on the weight of permeate. About 50% of the EC-treated water was collected as permeate. After each experiment, the membrane was cleaned by pumping hot DI water for 1 h prior to starting a new experiment.
2.4. MD Testing
The MD system used here is shown schematically in
Figure 3. A custom-made acrylic module was used. The total membrane surface area was 40 cm
2. The flow channel was 2 mm deep. A commercial 0.65 μm pore size PVDF membrane (Millipore, Billerica, MA, USA) was used. PTFE spacers were also used (ET 8700, Industrial Netting, Minneapolis, MN, USA) to provide mechanical support and promote mixing. Peristaltic pumps (Masterflex I/P, Cole Parmer, Vernon Hills, IL, USA) were utilized to pump the feed and permeate streams at 0.5 L/min on opposite sides of the membrane. The temperature of the permeate and feed tanks was maintained at 20 °C and 60 °C using an external chiller and heater, respectively (PolyScience, Niles, IL, USA). Experiments were run for about 8 h. We aimed for about 40% water recovery.
2.5. MDC Testing
MDC experiments were conducted in four stages. Initially, MD was run till about 10% of the feed was recovered in the permeate (about 115 min operation). The feed tank was then placed in a water bath containing ice that was constantly replaced. After about 15 min, the temperature of the feed reached 20 °C. It was then kept in the water bath for an additional 5 min. After removal of the precipitate, the feed tank was returned to the MD system, and MD recommenced once the temperature reached 60 °C. The precipitate was recovered and analyzed for the feasibility of mineral recovery. Precipitation occurs in the feed tank rather than on the membrane, increasing water recovery.
Using the weight change of the permeate tank, the water flux was calculated and normalized using the average flux during the first 15 min of operation. The permeate conductivity was measured using a conductivity meter (VWR, Radnor, PA, USA). Each MD and MDC experiment was conducted using 500 mL of PW with no pretreatment or with PW pretreated using EC UF.
An experiment was also conducted where the membrane was regenerated and reused. A membrane regeneration cycle was applied once 40% of the feed volume was recovered or there was no permeate weight increase for 20 min. During regeneration, DI water was pumped on both sides of the membrane at 0.5 L/min for 1 h.
4. Conclusions
This investigation is one of the first studies to investigate the use of EC UF MDC for treating hypersaline hydraulic fracturing PW. The combined EC UF MDC process was used to treat hydraulic fracturing PW. The PW manifested a high TDS, TSS, and TOC. Nevertheless, 40% of the feed volume was recovered. It is likely that greater water recovery is possible. By using crystallization after MD, precipitation on the membrane is suppressed. Adequate reduction in the PW TOC can be achieved using EC. UF is then used to efficiently remove the particulate matter produced during EC. The stability of the membrane is critical. Here, a commercially available PVDF membrane was used. The membrane was robust and easily regenerated.
The EC UF MDC technology can have an impact on water resources and oil and gas companies, as surface and groundwater form about 80% of the water used in hydraulic fracturing. The process developed here could be used to treat and reuse PW. If a crystallization step is not included, the operational cost of the process will be reduced while still providing significant water recovery. The possibility of mining the precipitate from the crystallization tank could lead to valuable byproducts that could help offset the cost of water treatment. The data collected from treating PW can be used to evaluate the integrated EC UF MDC process, which can guide further development of the process.