Pilot Demonstration of Reclaiming Municipal Wastewater for Irrigation Using Electrodialysis Reversal: Effect of Operational Parameters on Water Quality

Home , Electrodialysis reversal, Hydration number

membranes

Article Pilot Demonstration of Reclaiming Municipal Wastewater for Irrigation Using Electrodialysis Reversal: Effect of Operational Parameters on Water Quality

Xuesong Xu 1 , Qun He 2, Guanyu Ma 1, Huiyao Wang 1, Nagamany Nirmalakhandan 1 and Pei Xu 1,*

1 Department of Civil Engineering, New Mexico State University, Las Cruces, NM 88003, USA; [email protected] (X.X.); [email protected] (G.M.); [email protected] (H.W.); [email protected] (N.N.) 2 Carollo Engineers, Phoenix, AZ 85034, USA; [email protected] * Correspondence: [email protected]; Tel.: +1-575-646-5870

Abstract: The modification of ion composition is important to meet product water quality requirements, such as adjusting the sodium adsorption ratio of reclaimed water for irrigation. Bench- and pilot-scale experiments were conducted using an electrodialysis reversal (EDR) system with Ionics normal grade ion-exchange membranes (CR67 and AR204) to treat the reclaimed water in the Scottsdale Water Campus, Arizona. The goal is to investigate the impact of operating conditions on improving reclaimed water quality for irrigation and stream flow augmentation. The desalting efficiency, expressed as electrical conductivity (EC) reduction, was highly comparable at the same current density between the bench- and pilot-scale EDR systems, proportional to the ratio of residence time in the electrodialysis stack. The salt flux was primarily affected by the current density independent of flow rate, which is associated with linear velocity, boundary layer condition, and Citation: Xu, X.; He, Q.; Ma, G.; residence time. Monovalent-selectivity in terms of equivalent removal of divalent ions (Ca2+, Mg2+, Wang, H.; Nirmalakhandan, N.; Xu, P. and SO 2−) over monovalent ions (Na+, Cl−) was dominantly affected by both current density and Pilot Demonstration of Reclaiming 4 water recovery. The techno-economic modeling indicated that EDR treatment of reclaimed water Municipal Wastewater for Irrigation Using Electrodialysis Reversal: Effect is more cost-effective than the existing ultrafiltration/reverse osmosis (UF/RO) process in terms of Operational Parameters on Water of unit operation and maintenance cost and total life cycle cost. The EDR system could achieve Quality. Membranes 2021, 11, 333. 92–93% overall water recovery compared to 88% water recovery of the UF/RO system. In summary, https://doi.org/10.3390/ electrodialysis is demonstrated as a technically feasible and cost viable alternative to treat reclaimed membranes11050333 water for irrigation and streamflow augmentation.

Academic Editor: Xin Liu Keywords: electrodialysis; reclaimed water; desalination; ion-exchange membranes; ion permselectivity; irrigation Received: 9 April 2021 Accepted: 28 April 2021 Published: 30 April 2021 1. Introduction Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in The development of alternative water supplies such as municipal wastewater has been published maps and institutional affil- a viable solution to address the challenges of water scarcity and impaired water quality that iations. have influenced human activities and the environment [1–3]. Reclaimed water has been used for irrigation of turf and crops [4,5], groundwater recharge [6,7], potable reuse [8], cooling water for power plants [9], and other municipal and industrial applications [10]. The total dissolved solids (TDS) in such reclaimed water sources (usually above 1000 mg/L) and elevated concentrations of specific ions (e.g., sodium, calcium, chloride, nitrate, and Copyright: © 2021 by the authors. sulfate) can adversely impact potential beneficial use [11,12]. For instance, the use of Licensee MDPI, Basel, Switzerland. reclaimed water containing high sodium concentration for agricultural and golf course This article is an open access article distributed under the terms and irrigation would degrade soil condition [13,14], while the discharge of high chloride water conditions of the Creative Commons may jeopardize surface water discharge permit compliance due to failure of the Whole Attribution (CC BY) license (https:// Effluent Toxicity (WET) test [15]. creativecommons.org/licenses/by/ Reverse osmosis (RO) is currently the primary membrane separation process for 4.0/). water desalination and reuse [16,17]. However, the application of RO is hindered by

Membranes 2021, 11, 333. https://doi.org/10.3390/membranes11050333 https://www.mdpi.com/journal/membranes Membranes 2021, 11, 333 2 of 20

high energy demand, membrane scaling and fouling, and limited options for concentrate disposal [18]. Electrodialysis or electrodialysis reversal (EDR) that utilizes an electrical potential difference between the electrodes and ion-exchange membranes (IEMs) to remove salt ions provides an alternative to desalinate brackish water and wastewater [19,20]. The tendency of membrane scaling and fouling in electrodialysis can be significantly reduced because of paralleling water flow paths along the IEM surface (not passing through membranes) and relatively loose scaling/fouling layer due to low hydraulic pressure applied [21]. EDR enables the same electrochemical separation process as electrodialysis, except periodic reversal of the electrical polarity of the electrodes to inhibit deposition of potential scaling/fouling on the IEM surface [22,23]. In addition, electrodialysis barely rejects neutral species such as silica and organic matter, thereby further reducing the fouling and scaling propensity. Unlike RO that produces high purity permeate and requires post- treatment and stabilization to generate final product water, electrodialysis can flexibly tailor the desalination level and control ion composition in product water by adjusting operating parameters such as applied voltage, flow rate, and staging of the electrodialysis stacks. The electrodialysis process can directly produce product water with the desired water quality, thus avoiding the complexity of stabilizing the product water after desalination [24–26]. Therefore, electrodialysis could be a more cost-effective and energy-saving process than RO for desalination of impaired water sources for non-potable water reuse applications. Electrodialysis has been a commercial technology for decades, primarily for groundwater desalination. There is an increasing interest in electrodialysis for various applications including irrigation because of its selective separation capacity. More investigation of wastewater treatment using electrodialysis is needed to explore the transport behaviors of ions in multi-component solutions due to the complex interactions of ions in the solutions and competing transport of different ions through IEMs simultaneously [27]. Currently, most electrodialysis studies in treating reclaimed water were conducted in laboratory-scale systems with synthetic or real solutions [28–32]. It is critical to treat wastewater at scale- up electrodialysis systems for evaluating the feasibility and economic viability through long-term near full-scale operations, and bridging bench- and pilot-scale experiments to predicting commercial implementation [33,34]. In addition, the impacts of operational parameters need to be thoroughly analyzed at large-scale electrodialysis system for system design and operation. The applied voltage and current density (Equation (3) in Supplemental Information) are of importance for the overall performance of electrodialysis system [35]. The flow rate, associated with linear velocity, is also a key operating parameter to determine overall desalination and energy efficiencies because it affects the degree of mixing and controls the residence time of the desalting water inside the electrodialysis stacks [36,37]. Therefore, the net influence of flow rate on the salt transfer rate merits further investigation for reclaiming wastewater. As scaling can pose a challenge for treating wastewater using electrodialysis at high water recovery [38,39], the impact of hydraulic water recovery is also worth being evaluated under various operating conditions. This study investigated the application of electrodialysis to treat reclaimed water for golf course irrigation and instream flow augmentation. A bench-scale electrodialysis unit and a 2 electrical stage pilot-scale EDR system manufactured by the Suez Water Technologies & Solutions were operated continuously at an advanced wastewater treatment facility in Scottsdale, Arizona, USA, to investigate the technical feasibility and economic viability of the electrodialysis process for treating reclaimed water under different operating conditions. The reclaimed water needs to meet a sodium goal of 110 mg/L for golf course irrigation and a chloride goal of 150 mg/L to reduce the risks of WET failure for instream flow augmentation. The City of Scottsdale utilizes ultrafiltration (UF) as pretreatment and subsequent RO to desalinate a partial stream of the reclaimed water and blend it with the remaining stream. This approach requires subsequent blending and stabilization for final product water and generates a RO concentrate that requires proper disposal. Membranes 2021, 11, 333 3 of 20

The present study aims to evaluate the use of EDR as an alternative to UF/RO for treating the reclaimed water for non-potable water reuse, e.g., irrigation and in-stream augmentation. The main objectives of the study include the following: (1) assess the sodium and chloride removal behavior in treating the reclaimed water to meet the target water quality requirements for golf course irrigation and surface water discharge; (2) evaluate the impacts of pilot-scale EDR conﬁguration (e.g., number of staging and cell pairs) and operating conditions (e.g., applied voltage, ﬂow rate, and water recovery) on desalination performance and ion composition of the product water; (3) compare the desalting performance between bench- and pilot-scale electrodialysis for system scale- up and optimization; and (4) conduct a techno-economic analysis of the EDR process in comparison to the existing UF/RO system.

2. Materials and Methods 2.1. Water Quality and Analysis The bench- and pilot-scale experiments were conducted at the Scottsdale Water Cam- pus, Arizona, treating the denitrified tertiary effluent from the wastewater treatment plant (WWTP). During the testing period, the TDS concentration of the reclaimed water was 1131 ± 34 mg/L with an electrical conductivity of 1812 ± 61 µS/cm. The major ions in the reclaimed water included Na+ (246 ± 30 mg/L), Ca2+ (85 ± 8 mg/L), Mg2+ + − 2− (29 ± 1.4 mg/L), K (23 ± 1.2 mg/L), Cl (350 ± 18 mg/L), SO4 (253 ± 32 mg/L), and SiO2 (12.5 ± 1.1 mg/L). The relatively high concentrations of sodium and chloride make the reclaimed water unsuitable for beneficial reuse, specifically for irrigation in this case. The concentrations of other minor ions such as arsenic, phosphate, iron, manganese, nitrite, nitrate, bromide, and fluoride were negligible to impact the desalination performance. The pH of the reclaimed water was 7.4 ± 0.1, and the alkalinity was 155 ± 20 mg/L as CaCO3. The total organic carbon (TOC) concentration in the reclaimed water was 7.7 ± 0.15 mg/L. The electrical conductivity and pH of the water samples were measured using a conductivity and pH meter (Model 431-61, Cole-Parmer, Vernon Hills, IL, USA). The TOC was determined by a carbon analyzer (Shimadzu TOC-L, Kyoto, Japan), and major ions were measured using an ion chromatograph (IC, ICS-2100, Dionex, Sunnyvale, CA, USA). The concentrations of trace metallic elements were quantified using inductively coupled plasma mass spectrometry (ICP-MS, Elan DRC-e, PerkinElmer, Waltham, MA, USA). Alkalinity was measured using a digital titrator (Hach, Loveland, CO, USA) and 1.6 N sulfuric acid standard solution. The TDS concentration was measured following the evaporation method at 180 ◦C after filtering the water samples using a 0.45 µm cellulose acetate membrane filter (Toyo Roshi Kaisha, Ltd., Japan).

2.2. Bench- and Pilot-Scale Electrodialysis Systems The mobile, trailer-mounted EDR system (AQ3-1-4, Suez Wvs. ater Technologies & Solutions, Guelph, Ontario, Canada) is a pilot version of a commercial full-scale water treatment plant (Figure1). The pilot-scale EDR system can generate 3 to 13 gallons per minute (gpm, or 11.3 to 49 L per minute, L/min) product water. The Ionics normal grade IEMs (AR204-SZRA-412 and CR67-HMR-412, Suez Water Technologies & Solutions, Guelph, Ontario, Canada) were used for the bench- and pilot-scale electrodialysis testing. The thickness of the membranes is 0.6 mm for cation-exchange membranes CR67 and 0.5 mm for anion-exchange membranes AR204. Both IEMs have the same water content of 46%. The ion exchange capacity (IEC) of AR204 was higher than the CR67 (2.4 vs. 2.1 meq/g dry membrane), while the area electrical resistance of the AR204 measured in 0.01 N NaCl solution was lower than the CR67 (8 vs. 12 ohm–cm2). The patented mesh spacers (Mark IV-2) have a thickness of 0.076 cm and provide ﬂow channel and mixing for each stream in the EDR unit [40]. Platinum plated titanium metal plates were used as electrodes to apply electrical potential to the stack. One hundred cell pairs of CR67 and AR204, with effective area approximately 3200 cm2 (3.44 ft2) for each membrane, were Membranes 2021, 11, x 4 of 20

Membranes 2021, 11, 333 4 of 20 to apply electrical potential to the stack. One hundred cell pairs of CR67 and AR204, with effective area approximately 3200 cm2 (3.44 ft2) for each membrane, were installed in the EDR stack with 2 electrical stage and 2 hydraulic stage. Each stage had 50 pairs of mem- installedbranes. An in the independen EDR stackt voltage with 2 electricalsource was stage applied and to 2 hydrauliceach electrical stage. stage. Each stage had 50 pairs of membranes. An independent voltage source was applied to each electrical stage.

Reclaimed water

Electrode waste Electrode feed

Diluate Feed in Off-spec Off-spec

Concentrate make-up

Recycle Blowdown Total Waste

Pilot Testing Bench Testing

Figure 1. Schematic flow diagram of the electrodialysis systems. Figure 1. Schematic flow diagram of the electrodialysis systems. An in-line 10 µm cartridge filter was installed as a pretreatment step before the An in-line 10 µm cartridge filter was installed as a pretreatment step before the re- reclaimed water entering the electrodialysis stack. Feedwater was split into three streams: claimed water entering the electrodialysis stack. Feedwater was split into three streams: (I) a stream of feedwater (feed in) was progressively demineralized as desalted water (I) a stream of feedwater (feed in) was progressively demineralized as desalted water (product water or diluate); (II) a stream of feedwater was used as concentrate make-up (product water or diluate); (II) a stream of feedwater was used as concentrate make-up to to be mixed with concentrate recycled as the inlet concentrate; and (III) a small stream of feedwaterbe mixed (<4%) with wasconcentrate diverted recycled into the electrodeas the inlet rinsing concentrate; stream. Aand blower (III) a was small installed stream to of removefeedwater the chlorine(<4%) was gas diverted generated into in the the electrod electrodee rinsing rinse waste stream. stream. A blower The electrodewas installed rinse to remove the chlorine gas generated in the electrode rinse waste stream. The electrode rinse solution can be replaced by another source of aqueous solution (e.g., Na2SO4 solution) to avoidsolution the generationcan be replaced of chlorine by another gas. source of aqueous solution (e.g., Na2SO4 solution) to avoidThe the pilot-scale generation EDR of stackchlorine reversed gas. electrical polarity automatically every 15 min. The currentThe in thepilot-scale positive EDR polarity stack mode reversed was slightlyelectrical smaller polarity than automatically that in the negative every 15polarity min. The mode,current which in the had positive relatively polarity lower mode value was of electricalslightly smaller conductivity than that inproduct in the negative water. Allpolarity the reportedmode, which data herein had relatively were from lower the watervalue samplesof electrical obtained conductivity during thein product positive water. polarity All mode,the reported which had data similar herein water were from quality the as water in the samples negative obtained polarity. during Within the the positive first minute polar- afterity mode, polarity which change, had foulantssimilar water and scalesquality were as in flushed the negative off from polarity. the IEMs Within surfaces, the first and mi- thenute product after polarity water (namely change, off-spec foulants water) and scal couldes were not meet flushed the off required from the water IEMs quality. surfaces, A real-timeand the product conductivity water sensor (namely was off-spec installed water) in thecould diluate not meet stream the outletrequired to monitorwater quality. the waterA real-time quality andconductivity control a diversionsensor was valve installed to steer in thethe off-specdiluate waterstream into outlet the wasteto monitor stream the whenwater its quality TDS exceeded and control the product a diversion water qualityvalve to control steer the level. off-spec Both off-spec water andinto electrodethe waste rinsingstream water when were its TDS diverted exceeded into thethe wasteproduct stream, water which quality can control be recycled level. toBoth the off-spec feedwater and tankelectrode for further rinsing enhancing water were water diverted recovery into in th thee waste EDR system.stream, which can be recycled to the feedwaterDifferent tank water for recoveriesfurther enhancing (Equation water (2) in recovery Supplemental in the EDR Information) system. were achieved by adjusting the flow rates of the concentrate blowdown, feed in, and concentrate make-up. For example, to achieve an overall 50% water recovery, the total inlet feedwater was 14 gpm (53 L/min) to produce 7.5 gpm (28.4 L/min) product water. Considering the water loss

Membranes 2021, 11, 333 5 of 20

as off-spec water (~7% of product water), electrode rinsing waste 0.53 gpm (2 L/min), concentrate make-up stream 5.8 gpm (22.0 L/min), the overall water recovery was 50%. Operating parameters, including applied voltage, current density, pH, electrical conductivity, pressure, and flow rate of each water stream, were recorded by the supervisory control and data acquisition system (SCADA, Allen-Bradley, Milwaukee, WI, USA). The polyphosphate antiscalant (Hypersperse, MDC714, Suez Water Technologies & Solutions) with injection dosage (19 mg/L) calculated by the predictive WATSYSTM program was pumped continuously into the concentrate stream to inhibit scale formation/precipitation on the IEM surface. The WATSYSTM program was developed by Suez Water Technologies & Solutions for designing full-scale EDR systems. A bench-scale electrodialysis system with the same flow design as the pilot unit was operated independently alongside the pilot-scale EDR system to optimize the desalination performance and scale-up from the bench- to pilot-scale systems. The bench-scale laboratory unit included eleven pieces of CR67 and ten pieces of AR204 membranes, each with 220 cm2 (0.24 ft2) effective membrane area. The variable-flow micro-pumps (EW-75211-10, Cole-Parmer, Vernon Hills, IL, USA) were chosen to provide desired flow rates and pressure while minimizing pressure fluctuation inside the bench-scale electrodialysis stack. The two electrodialysis systems had the same stack design, configuration of spacers, and IEMs, which made a highly defensible comparison between the bench- and pilot-scale experiments possible. Before each experiment, all new membranes were equilibrated with the reclaimed water for at least 24 h. After the three-month pilot-scale testing, the membranes were removed from the pilot-scale EDR stack for membrane autopsy to identify possible fouling/scaling.

3. Results and Discussion 3.1. LCD Measurement Limiting current density (LCD) was studied to evaluate the concentration polarization (CP) in the electrodialysis system. Due to higher ion mobility through membrane matrix than that in solution, concentration polarization, also referred to as concentration gradient, was developed at the interface between membrane and solution. With an increase of the applied voltage, ion concentration on the diluate side would approach zero at the interface, and the current per unit effective membrane area at this point was known as LCD [41]. Water splitting occurs when the current density exceeds the LCD, resulting in reduced desalination efficiency by diverting electric energy to produce H+ and OH− from water electrolysis and contributing little to ion transport during desalination [42]. Most electrodialysis and EDR systems are commonly operated at a sub-LCD to achieve high energy efficiency [22], making the LCD an essential parameter in the electrodialysis process. During the bench- and pilot-scale electrodialysis experiments, the applied voltage was increased discretely every ten to fifteen minutes. LCD values were estimated based on the voltage–current response after a steady-state operational condition was reached. Two current–voltage characteristics were adopted to determine the LCD value (Figure2). The point that the curve slope changes in the voltage–current (V–I) curve indicated the LCD location in Figure2a,c. Sometimes it is impractical to identify the slope-changing point in the V–I curve, especially during desalination of high salinity water where consistent linearity of V–I relationship could be observed over a broad applied voltage range [43]. Cowan and Brown [44] developed the well-accepted method of plotting V/I against the 1/I curve, where the point with the minimum value of V/I was determined as the location of LCD (Figure2b,d). Membranes 2021, 11, x 6 of 20

The LCD value of the bench-scale unit was identified to be approximately 12 mA/cm2 (Figure 2a). The slope-turning point of the V/I–1/I curve in bench-scale electrodialysis was located at 135 Ω–cm2/cell-pair (Figure 2b), in accordance with the observation in Figure 2a. For the 1st electrical stage of the pilot-scale EDR, the LCD was determined to be 3.0 mA/cm2 (Figure 2c), while the estimation of the normalized electrical resistance at the LCD was close to 160 Ω–cm2/cell-pair (Figure 2d). The ohmic regime below the LCD had a highly linear correlation of current density versus applied voltage in both the bench- and pilot-scale electrodialysis stack (R2 > 0.99), revealing nearly constant electrical resistance during desalination within this current density range. Significant electrical resistance jump was encountered above the LCD point due to the depleting ion concentration at the membrane-solution interface, indicating lower energy efficiency compared to that in the ohmic regime. The data for LCD determination were collected during the positive polarity mode. The LCD measured in the negative polarity exhibited a similar trend but slightly lower by 8% (2.75 mA/cm2) than the data measured in the positive polarity mode. The 2nd electrical stage of the pilot-scale EDR exhibited different V–I and V/I–1/I behaviors (Figure 2e,f) compared to the 1st electrical stage. A lower minimum point located at current density 1.25 mA/cm2 with higher normalized electrical resistance 220 Ω–cm2/cell-pair was detected, due partly to the fact that the inter-stage diluate out, product water of the 1st electrical stage, was the feedwater of the 2nd electrical stage. LCD was reported to be proportional to feedwater salt concentration [43]. The non-linear V–I curve in the 2nd electrical stage (Figure 2e) was mainly caused by the varying electrical resistance associated with the changing salt concentration in its feedwater [45]. Similar trends were observed in our previous experiments of treating brackish groundwater using the same pilot-scale EDR stack at the Kay Bailey Hutchison Desalination Plant in El Paso, Texas [46]. For the 1st electrical stage, the pH of the product water decreased from 7.4 (same as the feedwater) to 6.4 with the increase of current density (0.3–4.0 mA/cm2) in Figure 3, Membranes 2021, 11, 333 indicating the water dissociation at higher current density [47], while in the concentrate6 of 20 stream, the pH remained relatively constant at 7.3. The same pH reduction trend was also observed in the negative polarity mode.

20 300 y = 3.7258x + 6.4253

) 16 R² = 0.9768 250 2

200 12 150 /cell-pair) 8 2 y = 8.6569x − 1.8811 100

4 R² = 0.9991 (Ω–cm 50 Voltage/current density Current density Current density (mA/cm 0 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.2 0.4 0.6 0.8 1.0 Membranes 2021, 11, x Voltage (volts/cell-pair) Reciprocal current density (cm2/mA) 7 of 20

(a) (b)

5 200 ) 2 4 y = 4.4298x + 0.8168 190 R² = 0.9866

3 180 /cell-pair) 2 2 170 (Ω–cm 1 y = 6.696x − 0.3054 160 Voltage/current density Current density Current density (mA/cm R² = 0.9976

0 150 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 1.0 Voltage (volts/cell-pair) Reciprocal current density (cm2/mA)

1.0 300 /cell-pair) 2

0.5 250 (Ωc–m Current density Current density (mA/cm Voltage/current density

0.0 200 0.0 0.2 0.4 0.6 0.8 0.0 1.0 2.0 3.0 4.0 Voltage (volts/cell-pair) Reciprocal current density (cm2/mA)

(e) (f)

Figure 2. Determination of limiting current densitydensity duringduring electrodialysiselectrodialysis ofof reclaimedreclaimed water.water. (a(a)) and and ( b(b)) V–I V–I and and V/I–1/I V/I–1/I st curves of the the bench-scale bench-scale electrodialysis electrodialysis (ED) (ED) system; system; (c ()c )and and (d (d) )V–I V–I and and V/I–1/I V/I–1/I curves curves of the of the1 electrical 1st electrical stage stage in the in nd pilot-scalethe pilot-scale electrodialysis electrodialysis reversal reversal (EDR (EDR) system; system; (e) and (e) and(f) V–I (f) and V–I andV/I–1/I V/I–1/I curves curves of the of 2 the electrical 2nd electrical stage stagein the in EDR the system. EDR system.

7.6

7.4

7.2

7.0

pH 6.8

6.6

6.4

6.2 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

2 Current density (mA/cm ) Figure 3. Product water pH as a function of the current density of the pilot-scale EDR system during electrodialysis of the reclaimed water.

Membranes 2021, 11, x 7 of 20

5 200 ) 2 y = 4.4298x + 0.8168 190 Membranes 2021 11 4 , , 333 R² = 0.9866 7 of 20

3 180 /cell-pair)

2 2 2 The LCD value of the bench-scale170 unit was identiﬁed to be approximately 12 mA/cm (Figure2a). The slope-turning point of the V/I–1/I curve in bench-scale electrodialysis 2 was located at 135 Ω–cm /cell-pair(Ω–cm (Figure2b), in accordance with the observation in 1 y = 6.696x − 0.3054 160 Voltage/current density Current density Current density (mA/cm FigureR² =2 a0.9976. For the 1st electrical stage of the pilot-scale EDR, the LCD was determined to be 3.0 mA/cm2 (Figure2c), while the estimation of the normalized electrical resistance at the 0 150 LCD was close to 160 Ω–cm2/cell-pair (Figure2d). The ohmic regime below the LCD had 0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8 1.0 a highly linear correlation of current density vs. applied voltage in both the bench- and Voltage (volts/cell-pair) Reciprocal current density (cm2/mA) pilot-scale electrodialysis stack (R2 > 0.99), revealing nearly constant electrical resistance

during desalination within this current density range. Signiﬁcant electrical resistance (c) (d) jump was encountered above the LCD point due to the depleting ion concentration at the 400 2.0 membrane-solution interface, indicating lower energy efﬁciency compared to that in the

) ohmic regime. The data for LCD determination were collected during the positive polarity 2 1.5 mode. The LCD measured in the negative350 polarity exhibited a similar trend but slightly lower by 8% (2.75 mA/cm2) than the data measured in the positive polarity mode. The 2nd electrical stage of the pilot-scale EDR exhibited different V–I and V/I–1/I behaviors 1.0 (Figure2e,f) compared to the 1 st electrical300 stage. A lower minimum point located at current /cell-pair) density 1.25 mA/cm2 with higher normalized2 electrical resistance 220 Ω–cm2/cell-pair was detected, due partly to the fact that the inter-stage diluate out, product water of the 0.5 250 1st electrical stage, was the feedwater(Ωc–m of the 2nd electrical stage. LCD was reported to Current density Current density (mA/cm be proportional to feedwater salt concentrationVoltage/current density [43]. The non-linear V–I curve in the 2nd 0.0 electrical stage (Figure2e) was mainly caused200 by the varying electrical resistance associated 0.0 0.2with the 0.4 changing 0.6 salt concentration 0.8 in its0.0 feedwater 1.0 [45]. 2.0 Similar trends 3.0 were 4.0 observed Voltagein our (volts/cell-pair) previous experiments of treating brackishReciprocal groundwater current density using (cm the2/mA) same pilot-scale EDR stack at the Kay Bailey Hutchison Desalination Plant in El Paso, Texas [46]. st (eFor) the 1 electrical stage, the pH of the product water(f) decreased from 7.4 (same as the feedwater) to 6.4 with the increase of current density (0.3–4.0 mA/cm2) in Figure3, Figure 2. Determination of limiting current density during electrodialysis of reclaimed water. (a) and (b) V–I and V/I–1/I indicating the water dissociation at higher current density [47], while in the concentrate curves of the bench-scale electrodialysis (ED) system; (c) and (d) V–I and V/I–1/I curves of the 1st electrical stage in the stream, the pH remained relatively constant at 7.3. The same pH reduction trend was also pilot-scale electrodialysis reversal (EDR system; (e) and (f) V–I and V/I–1/I curves of the 2nd electrical stage in the EDR system. observed in the negative polarity mode.

7.6

7.4

7.2

7.0

pH 6.8

6.6

6.4

6.2 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

2 Current density (mA/cm )

FigureFigure 3. 3.Product Product water water pH pH as as a a function function of of the the current current density density of of the the pilot-scale pilot-scale EDR EDR system system during during electrodialysis of the reclaimed water. electrodialysis of the reclaimed water.

3.2. Effect of Flow Rate on Desalination Performance To study the influence of flow rate on desalination performance, bench-scale experiments were first conducted at a product water flow rate of 1 L/min and 1.5 L/min, corresponding to a linear velocity of 4.1 and 6.1 cm/s, respectively. The percentage of TDS reduction (Equation (1) in Supplemental Information) decreased with the increase of flow rate from 1 L/min and 1.5 L/min (hydraulic residence time 13 s and 9 s, respectively), Membranes 2021, 11, x 8 of 20

3.2. Effect of Flow Rate on Desalination Performance To study the influence of flow rate on desalination performance, bench-scale experi-

Membranes 2021, 11, 333 ments were first conducted at a product water flow rate of 1 L/min and 1.5 L/min, corre-8 of 20 sponding to a linear velocity of 4.1 and 6.1 cm/s, respectively. The percentage of TDS reduction (Equation (1) in Supplemental Information) decreased with the increase of flow rate from 1 L/min and 1.5 L/min (hydraulic residence time 13 s and 9 s, respectively), and higherand higher current current density density was was required required to toachieve achieve the the same same conductivity conductivity reduction reduction at at a higherhigher flow flow rate (Figure 44a).a). TheThe differencedifference ofof TDSTDS removalremoval efficiencyefficiency atat thethe samesame currentcurrent 2 densitydensity became prominent from <9% below the currentcurrent densitydensity ofof 1.51.5 mA/cmmA/cm2 to >19% 2 aboveabove the current density of 8.1 mA/cm2.. The The positive positive impact impact of of a a higher higher flow flow rate rate on on salt salt removal,removal, associated associated with with higher higher linear linear veloci velocity,ty, was attributed to the decreased resistance ofof the the boundary boundary layer layer by by suppressing suppressing the di diffusionffusion layer between solution and membrane surfacessurfaces at the sideside ofof diluatediluate streamsstreams [ 48[48]].. On On the the other other hand, hand, decreased decreased residence residence time time at ata highera higher flow flow rate rate posed posed a negative a negative effect effect on on salt salt removal removal due due to lowerto lower residence residence time time for forion ion separation separation by passingby passing through through IEMs IEMs within wi thethin electrodialysis the electrodialysis stack. stack. The overall The overall nega- negativetive impact impact of higher of higher flow flow rate wasrate observedwas observed at the at current the current density density below below approximately approxi- 2 mately5.6 mA/cm 5.6 mA/cm(Figure2 (Figure4b), consistent 4b), consistent with other with observations other observations in bench-scale in bench-scale electrodialysis electro- dialysisusing normal using gradenormal IEMs grade [46 IEMs,49,50 [46,49,50],], which demonstrated which demonstrated the greater the influence greater influence of residence of residencetime over time the change over the of change the boundary of the boundary layer conditions layer conditions at varying at linearvarying velocity. linear velocity.

80% Bench-scale Bench-scale y = 0.0746x + 0.0129 12000 R² = 0.998 60% 1.5 L/min 9000 1 L/min 1.5 L/min 1.0 L/min 40% –kWh) 2 6000

y = 0.051x − 0.0177 (g/m 20% R² = 0.9966 3000 Conductivity Conductivity reduction Normalized salt removal

0% 0 024681012 024681012

Current density (mA/cm2) Current density (mA/cm2)

(a) (b) Bench-scale 120

) 1.5 L/min 2 90 1 L/min

30 Salt flux (g/hr–m

0 024681012

Current density (mA/cm2)

(c) Figure 4. The effect of flow rate on desalination performance in the bench-scale electrodialysis system. (a) total dissolved Figure 4. The effect of flow rate on desalination performance in the bench-scale electrodialysis system. (a) total dis- solids (TDS reduction efficiency in terms of conductivity reduction; (b) normalized salt removal; and (c) salt flux through solved solids (TDS reduction efficiency in terms of conductivity reduction; (b) normalized salt removal; and (c) salt flux IEMs. through IEMs.

With increasing current density, the boundary layer became the limiting factor of

ion transport through IEMs due to depletion of ions and sharp concentration gradient within the boundary layer. At a current density greater than 6 mA/cm2, the signiﬁcance of positive impact at a higher ﬂow rate prevailed the negative effect of less residence Membranes 2021, 11, 333 9 of 20

time, resulting in slightly enhanced overall normalized salt removal (Equation (5) in Supplemental Information) at a higher flow rate (Figure4b). The normalized salt removal exhibited a continuously decreasing trend with increase of the current density at both flow rates, which indicated the higher energy consumption at higher desalination requirement. To predict the specific salt concentration in product water as a function of current density, the absolute ion transport rate (namely, salt flux, Equation (4) in Supplemental Information) from the diluate stream to the concentrate stream was calculated. Salt flux through IEMs under different flow rates is depicted in Figure4c. The identical linear correlation between salt flux and current density demonstrated salt flux in electrodialysis is independent of hydrodynamic conditions, including the variation of linear velocity and boundary layer condition. The salt flux is primarily proportional to the current density, the driving force of ion transport [51]. Therefore, the linear variation of the salt flux and electrical charge density can be utilized to predict mass transport under different operating conditions with the same IEMs and spacers installed in the electrodialysis stack. In the pilot-scale testing, a baseline flow rate, namely, 1× flow rate of 7.5 gpm (28.4 L/min product water), was chosen because of the same linear velocity as that in the bench-scale experiments at flow rate 1.5 L/min (a linear velocity of 6.1 cm/s). More- over, with 1× flow rate (baseline), 1.25× flow rate of 9.4 gpm (35.5 L/min product water), and 1.5× flow rate of 11.3 gpm (42.6 L/min product water), the flow rate conditions were investigated in the pilot-scale EDR system. The percent salt removal efficiency exhibited decreasing trends as the flow rate increased under the same current density (Figure5a). The normalized salt removal indicated an overall positive impact of a higher flow rate at the starting operating conditions (Figure5b). Unlike having a short residence time in the bench-scale testing (13 s vs. 9 s), the residence time was 60 s, 48 s, and 40 s corresponding to 1×, 1.25×, and 1.5× flow rate, respectively. Salt ions in the diluate stream had a relatively long enough contact time to transport through IEMs, minimizing the influence of residence time on salt removal. The impact of a higher flow rate with increasing linear velocity demonstrated a dominant positive impact at initial current density tested in the pilot-scale electrodialysis, contrary to the overall negative effect of increased flow rate on desalination performance at low current density in the bench-scale system. Salt flux as a function of current density exhibited a single linear trend, regardless of the different flow rates in the pilot-scale testing (Figure5c). The trend further confirmed the linear correlation of salt flux as a function of current density. During electrodialysis of the reclaimed water, the bench-scale system achieved higher normalized salt removal of 6.10 kg salt/m2–kWh at the current density of 2.6 mA/cm2, while the pilot-scale system only achieved 0.08 kg salt/m2–kWh at the current density of approximately 2.6 mA/cm2 under the same linear velocity of 6.1 cm/s. The difference of normalized salt removal between the bench- and pilot-scale electrodialysis systems is attributed to the electrodialysis stack configuration and setup. The bench-scale unit has a smaller effective membrane area and a shorted flow path than the pilot system. This short flow path, combined with short hydraulic retention time (HRT), resulted in relatively smaller conductivity change between influents and effluents of diluate and concentrate chambers and reduced water splitting at the end of diluate stream, thus increasing desalination performance. This phenomenon was reported by another study using a different electrodialysis system (PCCell 64 0 02, PCCell GmbH, Germany) with a 64 cm2 effective membrane area of the same membrane type but different spacers [52]. A significantly higher normalized salt removal was observed in comparison with this study in treating solution with similar TDS (~1100 mg/L) in bench-scale testing. This result was also confirmed in treating brackish groundwater, using the same bench- and pilot-scale electrodialysis systems as in this study [46]. Membranes 2021, 11, 333 10 of 20 Membranes 2021, 11, x 10 of 20

100% Pilot-scale 100 Pilot-scale

1x flow rate 80% 90 1.25x flow rate 60% 1.5x flow rate 80 –kWh) 2 40%

1x flow rate (g/m 1.25x flow rate 70

Conductivity Conductivity reduction 20%

1.5x flow rate Normalized salt removal 0% 60 2.0 2.2 2.4 2.6 2.8 3.0 3.2 2.0 2.2 2.4 2.6 2.8 3.0 3.2

Current density (mA/cm2) Current density (mA/cm2)

(a) (b)

50 Pilot-scale ) 2 40

20 Salt flux (g/hr–m 1x flow rate 10 1.25x flow rate 1.5x flow rate 0 2.0 2.2 2.4 2.6 2.8 3.0 3.2

Current density (mA/cm2)

(c)

Figure 5. The effect of flowflow raterate onon desalinationdesalination performance performance in in the the pilot-scale pilot-scale EDR EDR system. system. (a ()a TDS) TDS reduction reduction efficiency efficiency in in terms of conductivity reduction; (b) normalized salt removal; and (c) salt flux through ion exchange membranes (IEMs. terms of conductivity reduction; (b) normalized salt removal; and (c) salt flux through ion exchange membranes (IEMs).

AsAs linearity linearity of of salt salt removal removal efficiency efficiency and and salt salt flux flux found found in both in boththe bench- the bench- and pilot- and scalepilot-scale experiments, experiments, the overall the overall desalination desalination performance performance from fromthe bench- the bench- to pilot-scale to pilot- electrodialysisscale electrodialysis was expected was expected to scale-up to scale-up under under the same the samehydraulic hydraulic configurations configurations (e.g., linear(e.g., linearvelocity, velocity, spacer spacer pattern, pattern, and staging) and staging) [46]. With [46 the]. With same the linear same velocity linear of velocity 6.1 cm/s, of the6.1 cm/s,desalination the desalination performance performance of the bench-scale of the bench-scale and the 1st and electrical the 1st stageelectrical in the stage pilot- in scalethe pilot-scale electrodialysis electrodialysis is evaluated is evaluated in Figure in 6. Figure Generally,6. Generally, both systems both systems exhibited exhibited similar trendssimilar with trends increasing with increasing current currentdensity. density. High-linear High-linear correlations correlations between between percent percentsalt re- movalsalt removal efficiency efficiency and current and current density density (Figure (Figure 6a) were6a) observed were observed (R2 > 0.99) (R 2 in> 0.99)the bench- in the scalebench-scale and 1st andelectrical 1st electrical stage of stagethe pilot-scale of the pilot-scale system. system.The exhibited The exhibited overlapping overlapping salt flux trendssalt flux indicated trends indicatedthe nearly the identical nearly normal identicalized normalized desalting capacity desalting between capacity the between bench- andthe pilot-scale bench- and electrodialysis pilot-scale electrodialysis systems with scalable systems geometry with scalable and similar geometry hydrodynamics and similar (Figurehydrodynamics 6b). It represented (Figure6b ).a Ithigh represented similarity a highin desalination similarity inperf desalinationormance, which performance, can be utilizedwhich can to evaluate be utilized the tooverall evaluate desalination the overall of large-scale desalination electrodialysis of large-scale based electrodialysis on experi- mentsbased of on the experiments scale-down of bench-scale the scale-down system. bench-scale The desalination system. efficiency The desalination can be expressed efficiency as:can be expressed as: Conductivity reduction in pilot-scale system = α × Conductivity reduction in bench- scale unit × (Current Density pilot/Current Density bench) [46]. The ratio α was approximately 3.6 from Figure 6a, matching the ratio (3.4) of the residence time of pilot- over bench-scale experiments (HRT of 30 s over 9 s). The con- sistency of high similarity between bench- and pilot-scale testing treating reclaimed water

Membranes 2021, 11, x 11 of 20

Membranes 2021, 11, 333 (TDS~1150 mg/L, this study) and brackish groundwater (TDS~2700 mg/L, our11 previous of 20 study) presented a reasonable accuracy in simulating pilot-scale electrodialysis system using bench-scale testing results.

Bench-scale vs Pilot-scale Bench-scale vs Pilot-scale 80% 150 )

Bench Pilot 2 120 60% y = 0.1818x + 0.0085 R² = 0.9924 90 40% 60 Salt flux (g/hr-m

20% y = 0.051x - 0.0177 Bench Pilot R² = 0.9966 30 Conductivity Conductivity reduction

0% 0 024681012 024681012 Current density (mA/cm2) Current density (mA/cm2)

(a) (b)

FigureFigure 6. Scale-up 6. Scale-up of desalination of desalination performance performance from from the bench-the bench- to pilot-scale to pilot-scale electrodialysis electrodialysis of reclaimed of reclaimed water. water. (a) TDS(a) TDS removalremoval in terms in terms of conductivity of conductivity reduction; reduction; (b) overall (b) overall salt ﬂux.salt flux.

3.3.Conductivity Impact of Applied reduction Current in pilot-scale Density on system Desalination = α × andConductivity Ion Separation reduction in bench- scale unitConcentration× (Current Density polarization pilot/Current with increased Density current bench) [density46]. has a significant impact onThe the ratio desalinationα was approximately performance of 3.6 electrod from Figureialysis6 a,and matching EDR system the ratios [53]. (3.4) To ofdetermine the residencethe current time density of pilot- that over ensures bench-scale meeting experiments the required (HRT standards of 30 s over of target 9 s). Theions consis-(i.e., Na+< tency110 of mg/L, high similarityCl− < 150 betweenmg/L), the bench- pilot-scale and pilot-scale EDR system testing was treating operated reclaimed at several water current (TDS~1150densities mg/L, for treating this study) the reclaimed and brackish water groundwater at the baseline (TDS~2700 flow rate mg/L, of 7.5 gpm our previous (28.4 L/min study)product presented water, a or reasonable 6.1 cm/s). accuracyTable 1 summar in simulatingizes the pilot-scale corresponding electrodialysis parameters system of the 1st usingand bench-scale 2nd electrical testing stages results. in the EDR pilot-scale system.

3.3.Table Impact 1. ofSummary Applied of Current operational Density parameters on Desalination and data and in different Ion Separation cases. Concentration polarization with increased current density has a significant impact on the desalinationParameters performance Case#1(C#1) of electrodialysis Case#2(C#2) and EDR systems Case#3(C#3) [53]. To determine Case#4(C#4) the current densityElectrical that ensures Stage meeting 1st the required2nd standards1st 2nd of target1st ions (i.e.,2ndNa +< 1101st mg/L2nd, Cl− < 150Voltage, mg/L), (V) the pilot-scale 14 EDR13 system19 was operated19 at23 several 21 current 26 densities 24 for treatingCurrent, the reclaimed (A) water 4.7 at the3.6 baseline 6.7 flow rate4.8 of 7.58.3 gpm (28.44.8 L/min9.5 product 4.9 st water, orCurrent 6.1 cm/s). Density, Table 1 summarizes the corresponding parameters of the 1 and 2nd 1.5 1.1 2.1 1.5 2.6 1.5 3.0 1.5 electrical stages(mA/cm) in the2 EDR pilot-scale system. WithSalt increasingRemoval, (%) applied 29.6% voltage 22.3% and current 37.3% density,42.2% the46.9% salt removal48.1% efficiency55.5% 59.4% increased, coupling with the increased flux of different ions and reduced salt concentration Salt Removal Rate, in the product water. The overall571 percent302 salt719 removal510 efficiency904 increased492 1070 from 45%509 (g/hr) (Case#1) to 82% (Case#4). Meanwhile, the highest percent TDS reduction efficiency of Energy Consumption, any single hydraulic stage was65.8 below46.8 60%127.3 at the four91.2 cases193.2 tested, 98.7 within 244.4 the typical 117.6 optimal design(kWh) range of commercial EDR system [54]. The sodium concentration in the productEnergy water Consumption, decreased from 194 mg/L to 126 mg/L from Case#1 to Case#2. The chloride 0.12 0.15 0.18 0.18 0.21 0.20 0.23 0.23 concentration(kWh/kg dropped salt) to 101 mg/L in the product water of Case#2, meeting the required goal (<150Na mg/L+, (mg/L)), while the 244 sodium 194 concentration 204 was126 97 mg/L191 in Case#3. 97 169 Using the 75 TM correlationCl of−, (mg/L) current density 279 and salt205 concentration 204 in101 the WATSYS171 72program, 147 the goal 50 of sodium concentration 110 mg/L in the product water can be met at the current density 2 st 2 nd of 2.36 mA/cmWith increasingfor the 1 appliedelectrical voltage stage and and current 1.5 mA/cm density,for the the salt 2 removalelectrical efficiency stage. In in- contrast,creased, chloride coupling concentration with the increased in product flux water of different was significantly ions and reduced below the salt requirement concentration (<100in mg/L).the product Therefore, water. the The pilot-scale overall percent EDR system salt removal using the efficiency normal grade increased membranes from 45% can meet the sodium and chloride water quality goals of treating reclaimed water for golf course irrigation and surface discharge. The normalized salt mass removal exhibited a decreasing trend with the increase of current density, e.g., 0.12 kg salt/m2–kWh to 0.07 kg

Membranes 2021, 11, 333 12 of 20

salt/m2–kWh from Case#1 to Case#4, revealing reduced energy efﬁciency when operating at higher current density.

Table 1. Summary of operational parameters and data in different cases.

Parameters Case#1(C#1) Case#2(C#2) Case#3(C#3) Case#4(C#4) Electrical Stage 1st 2nd 1st 2nd 1st 2nd 1st 2nd Voltage, (V) 14 13 19 19 23 21 26 24 Current, (A) 4.7 3.6 6.7 4.8 8.3 4.8 9.5 4.9 Current Density, (mA/cm)2 1.5 1.1 2.1 1.5 2.6 1.5 3.0 1.5 Salt Removal, (%) 29.6% 22.3% 37.3% 42.2% 46.9% 48.1% 55.5% 59.4% Salt Removal Rate, (g/hr) 571 302 719 510 904 492 1070 509 Energy Consumption, (kWh) 65.8 46.8 127.3 91.2 193.2 98.7 244.4 117.6 Energy Consumption, (kWh/kg salt) 0.12 0.15 0.18 0.18 0.21 0.20 0.23 0.23 Na+, (mg/L) 244 194 204 126 191 97 169 75 Cl−, (mg/L) 279 205 204 101 171 72 147 50

The specific ion transport is depicted in Figure7. In Case#1, the flux of calcium was dominant, reaching 191 meq/hr-m2 compared to 139 meq/hr–m2 for sodium (Figure7a), even though the ratio of equivalent concentration of calcium over sodium was less than 0.4 in the feedwater. Selective transport of magnesium over sodium was also noticed, considering the lower ratio of equivalent concentration of magnesium over sodium in the feedwater (approximately 0.2). The selective transport of divalent cations (e.g., Ca2+ and Mg2+) over monovalent cation (Na+) was mainly attributed to the higher electrostatic attraction between the CEMs and the divalent ions with two charges, in comparison to monovalent cation with one charge [46]. For instance, the percent removal efficiency of calcium and magnesium reached 84% and 83% at Case#1 with low current density, respectively, while removal efficiency of 24% and 39% were recorded for sodium and potassium, respectively. 2− 2 Enhanced transport of divalent anion (SO4 ) was also detected (195 meq/hr–m ) over monovalent anion (262 meq/hr–m2 for Cl−) in Figure7b, in respective to less than 0.5 of equivalent concentration ratio of sulfate over chloride in the feedwater. With increasing current density, the increase of salt removal was mostly carried out by the transport of Membranes 2021, 11, x monovalent ions. The flux of monovalent ions increased linearly with the increase13 of of20

current density, while divalent ions displayed a non-linearity trend.

500 Cation transport 500 Anion transport 435

) 404 396 ) 2

Na K 2 Cl SO4 400 363 400 m 343 Mg Ca –

300 300 262 229 228 226 199 204 191 185 195 200 161 200 139 121 120

102 107 Ion flux (meq/hr Ion (meq/hr–m flux 100 100

13 18 24 25 0 0 C#1 C#2 C#3 C#4 C#1 C#2 C#3 C#4

(a) (b)

Figure 7. IonIon transporttransport inin the the pilot-scale pilot-scale EDR EDR for for treating treating the the reclaimed reclaimed water water at 50% at 50% water water recovery recovery and baselineand baseline flow rates.flow (rates.a) Ion (a flux) Ion of flux major of major cations; cations; and (b and) ion (b flux) ion of flux major of major anions. anions.

Selectivity of monovalent vs divalent ions 1000 Monovalent vs divalent ion transport 6 5.5 Monovalent ions

Ca/Na )

2 800 5 Divalent ions SO4/Cl 4 600 2.9 3 400 2.03 2.0 1.7 Ion flux (g/hr–m 2 1.47 1.28 1.12 200 1 Relative transport number (RTN) 0 0 C#1 C#2 C#3 C#4 C#1 C#2 C#3 C#4

(a) (b) Figure 8. Selectivity of monovalent vs. divalent ion (a) relative transport number of calcium over sodium and sulfate over chloride; and (b) ion flux of monovalent over divalent ions.

Comparison between the measured individual ion concentration data and the projection results using the WATSYSTM program was investigated (Figure S1 in Supplemental Information). The WATSYSTM program projected well the overall salt removal in terms of conductivity reduction and pH change. However, there were discrepancies between the measured ion concentrations and the projection results. The program also projected fairly well the transport of monovalent ions, while for divalent ions (e.g., Ca2+, Mg2+, and SO42−), the differences between projected and measured data were in the range of 10–40%. The monovalent ions demonstrated a fairly linear correlation between current density and ion flux, while non-linearity was noticed in divalent ions separation (Figure 8b). The nonlinear trend of divalent ions might contribute to the discrepancies between the measured and the WATSYSTM projected removal data. A low concentration of organic matter existed in the reclaimed water, measured as TOC of 7.7 ± 0.2 mg/L. The TOC concentration in product water ranged from 7.2 mg/L to 7.6 mg/L in the bench-scale electrodialysis system. One product water from Case#1 was measured as 5.03 mg/L of TOC. It was reported that organic matter in the reclaimed water is mostly neutral substances, which would not migrate under electric field [57]. Therefore,

Membranes 2021, 11, 333 13 of 20

Membranes 2021, 11, x Selective separation of divalent ions vs. monovalent ions was studied based on13 of Na 20

selectivity for cations and Cl selectivity for anions, deﬁned as relative transport number (RTN, Equation (6) in Supplemental Information). The RTN calculation is based on the widely-adopted equation developed by Sata et al. [55]. Na+ was speciﬁed as the reference − 500 Cationcation, transport and Cl was chosen to be the reference500 anion.Anion The RTN transport is measured as the amount of a given ion removed (in meq) divided by its arithmetic average equivalent435 concentration

) 404 396 ) 2 of initial and ﬁnal concentration in the diluate streams, as opposed to the amount of Na K 2 Cl SO4 400 363 400 m 343 Mg Careference ion removed (in meq) divided– by the average arithmetic equivalent the reference ion concentration of initial and ﬁnal concentration. 300 300 The trend of the RTN value of calcium was262 similar to that of magnesium, indicating 229 228 226 similar transport199 behavior for the major divalent cations.204 Divalent ions were selectively 191 185 195 200 transported161 over monovalent ions (RTN200 > 1) in the entire current density range tested. 139 121 120

102 The107 elevated removal of divalentIon flux (meq/hr ions was pronounced at low current density, i.e., the Ion (meq/hr–m flux 100 maximum RTN value of 5.5 for calcium100 (Figure8a). The downtrend of RTN for both 13 divalent18 cation24 and anion25 was attributed to the sharp concentration gradient in the bound- 0 ary layer as current density increased, enhancing0 the transport of monovalent ions in the C#1boundary C#2 layer C#3 because C#4 of higher diffusivity ofC#1 monovalent C#2 over C#3 divalent C#4 ions [56] and the downshifted degree of increment for divalent removal with the increase of current density (Figure(a) 7). Furthermore, the total flux of monovalent(b ions) increased linearly with the increase of current density, while the overall flux of divalent ions fluctuated in a relatively Figure 7. Ion transport in the pilot-scale EDR for treating the reclaimed2 water at 50% water recovery and baseline flow rates. (a) Ion flux of major cations;narrow and range (b) ion around flux of 500 major meq/hr–m anions. (Figure8b).

Selectivity of monovalent vs divalent ions 1000 Monovalent vs divalent ion transport 6 5.5 Monovalent ions

Ca/Na )

2 800 5 Divalent ions SO4/Cl 4 600 2.9 3 400 2.03 2.0 1.7 Ion flux (g/hr–m 2 1.47 1.28 1.12 200 1 Relative transport number (RTN) 0 0 C#1 C#2 C#3 C#4 C#1 C#2 C#3 C#4

(a) (b)

Figure 8. SelectivitySelectivity of monovalent vs. divalent divalent ion ion ( (aa)) relative relative transport transport number number of of calcium calcium over over sodium sodium and and sulfate sulfate over chloride; and (b) ion flux of monovalent over divalent ions. chloride; and (b) ion ﬂux of monovalent over divalent ions.

ComparisonComparison between thethe measuredmeasured individualindividual ion ion concentration concentration data data and and the the projec- pro- jectiontion results results using using the the WATSYS WATSYSTMTMprogram program was was investigated investigated (Figure(Figure S1S1 in Supplemental Information).Information). The The WATSYS WATSYSTM programprogram projected projected well well the the overall overall salt salt removal in terms of conductivityconductivity reduction reduction and and pH pH change. change. However, However, there were discrepancies between the measuredmeasured ion ion concentrations concentrations and the the projection projection results. The The program program also also projected projected fairly 2+2+ 2+2+ 22−− well the transport ofof monovalentmonovalent ions, ions, while while for for divalent divalent ions ions (e.g., (e.g., Ca Ca, Mg, Mg ,, and and SO SO44 ),), thethe differences differences between between projected projected and and measured measured data data were in the range of 10–40%.10–40%. The monovalentmonovalent ions ions demonstrated demonstrated a fairly fairly linear linear correlation correlation between current density and ion flux,ﬂux, while non-linearity waswas noticednoticedin in divalent divalent ions ions separation separation (Figure (Figure8b). 8b). The The nonlinear nonlin- eartrend trend of divalent of divalent ions mightions might contribute contribute to the to discrepancies the discrepancies between between the measured the measured and the andWATSYS the WATSYSTM projectedTM projected removal removal data. data. AA low low concentration of of organic organic matter matter existed existed in in the the reclaimed water, measured as TOCTOC of 7.7 ± 0.20.2 mg/L. mg/L. The The TOC TOC concentration concentration in in product product water ranged from 7.27.2 mg/Lmg/L to 7.6 mg/L in the bench-scale electrodialysis system. One product water from Case#1 was measured as 5.03 mg/L of TOC. It was reported that organic matter in the reclaimed water is mostly neutral substances, which would not migrate under electric field [57]. Therefore,

Membranes 2021, 11, x 14 of 20

organic removal was not further discussed in this study because there was no significant decrease of TOC in product water using electrodialysis.

3.4. Effect of Water Recovery Membranes 2021, 11, 333 The salt reduction decreased with the increase of water recovery in the pilot-scale14 of 20 EDR system due to a higher concentration gradient between the diluate and concentrate streams. The variation of conductivity reduction was less than 15% when the water recov- 7.6ery mg/Lincreased in the from bench-scale 50% to 80%. electrodialysis The concentration system. of One individual product ion water increased from Case#1 in product was measuredwater as the as 5.03water mg/L recovery of TOC. increased It was reported(Figure 9a). that The organic pH of matter product in the waters reclaimed at different water iswater mostly recovery neutral was substances, closely matching which would (pH not7.0–7 migrate.1). The under variations electric of ﬁeld percent [57]. removal Therefore, of organicthe major removal cations was were not within further 2% discussed as water inrecovery this study ranging because from there 50% was to no70%, signiﬁcant except a decreasesharp drop of TOCat water in product recovery water of 80% using (Figure electrodialysis. 9b). The transport of chloride was relatively more sensitive to water recovery compared to other major ions. The highest equivalent 3.4.concentration Effect of Water of chloride Recovery in concentrate might restrain the chloride transport from the diluateThe to salt concentrate reduction stream. decreased with the increase of water recovery in the pilot-scale EDRThe system selective due to separation a higher concentrationof divalent ions gradient versus betweenmonovalent the diluateions is presented and concentrate in Fig- streams.ure 10. For The cations, variation the RTN of conductivityvalue (2.8–3.0) reduction was relatively was lessstable than when 15% the when water the recovery water recoveryincreased increasedfrom 50% fromto 70%, 50% while to 80%.a sharp The increase concentration to 3.9 (~40% of individual increase) at ion water increased recovery in productof 80%. Stronger water as back the water diffusion recovery potential increased of monovalent (Figure9a). cations The pH from of productthe concentrate waters atstreams different contributed water recovery to the enhanced was closely divalent matching cation (pH selectivity 7.0–7.1). Theat higher variations water of recovery. percent removalHowever, of the the RTN major value cations of sulfate were within steadily 2% increased as water recoveryfrom 1.47 ranging to 1.99 as from water 50% recovery to 70%, exceptincreased a sharp from drop50% to at 70%, water then recovery jumped of to 80% 2.64 (Figure at water9b). recovery The transport of 80%. of chloride was relativelyAfter morefinishing sensitive all the to experiments, water recovery minor compared inorganic to scaling other majorwas spotted ions. Theon the highest CEM equivalentsurface, while concentration no scalants ofwere chloride observed in concentrate on the AEM might surface restrain by comparing the chloride the transportused and frompristine the membranes. diluate to concentrate stream.

1000 50% recovery 60% recovery 50% recovery 100% 70% recovery 80% recovery 60% recovery 70% recovery 80% 100 80% recovery 60%

10 40% Percent removal (%) Concentration (mg/L) 20%

1 0% Na K Ca Mg Cl SO4 Cond pH Na K Ca Mg Cl SO4 Cond pH

(a) (b)

Figure 9.9. SaltSalt rejectionrejection as a functionfunction of water recovery in treating reclaimed water (a) waterwater quality in product water; (b) percentagepercentage removalremoval ofof conductivityconductivity andand majormajor ions.ions.

The selective separation of divalent ions vs. monovalent ions is presented in Figure 10. For cations, the RTN value (2.8–3.0) was relatively stable when the water recovery increased from 50% to 70%, while a sharp increase to 3.9 (~40% increase) at water recovery of 80%. Stronger back diffusion potential of monovalent cations from the concentrate streams contributed to the enhanced divalent cation selectivity at higher water recovery. However, the RTN value of sulfate steadily increased from 1.47 to 1.99 as water recovery increased from 50% to 70%, then jumped to 2.64 at water recovery of 80%.

Membranes 2021, 11, 333 15 of 20 Membranes 2021, 11, x 15 of 20

6 Selectivity of monovalent vs divalent ions 800 Monovalent vs divalent ion flux Monovalent ions 5 700 )

2 Divalent ions Ca/Na SO4/Cl 3.9 600 4 500 2.9 3.0 2.8 3 2.64 400 1.99 1.70 300 2 1.47 200 1 ion Ion flux (g/hr–m 100 Relative transport number (RTN) 0 0 50% 60% 70% 80% 50% 60% 70% 80% Water recovery Water recovery

(a) (b)

Figure 10. Selectivity of monovalent versus vs. divalent divalent ion ion as a as function a function of water of water recovery. recovery. (a) Relative (a) Relative transport transport number; number; and and(b) ion (b) ﬂuxion flux of monovalent of monovalent over over divalent divalent ions. ions.

3.5. Techno-EconomicAfter finishing allAnalysis the experiments, minor inorganic scaling was spotted on the CEM surface,Based while on the no scalantstesting results, were observed alternative on thetreatment AEM surface configurations by comparing were developed the used and in comparisonpristine membranes. to the existing UF/RO system in Scottsdale Water Campus as baseline. The blending analysis (Table 2) and cost comparison (Table 3) were presented for a one million 3.5. Techno-Economic Analysis gallon per day (mgd, or 3785 m3/d) feed water flow on a realistic and common basis. Cost evaluationBased was on the performed testing results, in accordance alternative wi treatmentth the Association configurations for the were Advancement developed of in Costcomparison Engineering to the (AACE) existing [58]. UF/RO A detailed system cost in Scottsdalecomparison Water was developed Campus as to baseline. evaluate Thethe totalblending capital analysis cost, annual (Table2 )operation and cost comparisonand maintenance (Table 3(O&M)) were presentedcost, and for20-year a one life million cycle 3 costgallon for per the day proposed (mgd, ortreatment 3785 m /d)alternatives feed water (Detailed flow on aassumption realistic ands in common Supplemental basis. Cost In- formation).evaluation wasTables performed 2 and 3 summarize in accordance the calculated with the Association costs and system for the configuration Advancement and of performanceCost Engineering for the (AACE) UF/RO [ 58system]. A detailed as well costas the comparison 2-stage EDR was and developed 4-stage EDR to evaluate design basedthe total on WATSYS capital cost,TM simulation annual operation and pilot and testing maintenance results. (O&M) cost, and 20-year life cycle cost for the proposed treatment alternatives (Detailed assumptions in Supplemental TableInformation). 2. Blending Tables analysis2 and comparison3 summarize between the calculatedEDR and baseline. costs and system configuration and performance for the UF/RO system as well as the 2-stage EDR and 4-stage EDR design based on WATSYSTM simulation2-Stage and pilot EDR testing results. 4-Stage EDR Parameters Baseline (UF/RO) EDR EDR EDR (Testing) EDR (Testing) Table 2. Blending analysis comparison(WATSYS) between EDR and baseline.(WATSYS)

Feed Water Flow (mgd) 1 1 2-Stage 1 EDR 1 4-Stage EDR 1 Baseline Feed Water NaParameters (mg/L) 235 235 235 235 235 (UF/RO) EDR EDR EDR EDR % Flow Treated 60.5% 69.0%(WATSYS) 100.0%(Testing) 69.0%(WATSYS) 78.0%(Testing) Overall Recovery 88% 93% 92% 93% 92% Feed Water Flow (mgd) 1 1 1 1 1 UnitFeed Recovery Water Na (mg/L) 85% 235 90% 235 90% 235 90% 235 90% 235 Blended Water% FlowFlow Treated (mgd) 0.88 60.5% 0.93 69.0% 0.92 100.0% 0.93 69.0% 0.92 78.0% Product WaterOverall Na Recovery(mg/L) 110 88% 110 93% 129 92% 110 93% 110 92% Unit Recovery 85% 90% 90% 90% 90% ProductBlended TDS Water (mg/L) Flow (mgd) 530 0.88 522 0.93 489 0.92 522 0.93 433 0.92 ConcentrateProduct Flow Water (gpm) Na (mg/L) 60 110 48 110 69 129 48 110 54 110 ConcentrateProduct TDS TDS(mg/L) (mg/L) 7530 530 9662 522 7130 489 9662 522 9662 433 ConcentrateConcentrate Na (mg/L) Flow (gpm) 1524 60 2136 48 1715 69 1927 48 1715 54 Concentrate TDS (mg/L) 7530 9662 7130 9662 9662 NumberConcentrate of Product NaLine (mg/L) - 1524 7 2136 8 1715 7 1927 6 1715 NumberNumber of Stages of Product Line - - 2 7 2 8 4 7 4 6 Number of Stages - 2 2 4 4

Membranes 2021, 11, 333 16 of 20

Table 3. Cost comparison between EDR and baseline.

2-Stage EDR 4-Stage EDR Parameters Baseline (UF/RO) EDR (WATSYS) EDR (Testing) EDR (WATSYS) EDR (Testing) UF $599,000 $- $- $- $- Residuals Handling $765,000 $- $- $- $- RO $889,000 $- $- $- $- EDR $ - $1,820,000 $2,080,000 $2,520,000 $2,940,000 Building $652,000 $435,000 $435,000 $435,000 $435,000 Civil Site Works (5%) $146,000 $113,000 $126,000 $148,000 $169,000 Electrical and I&C (25%) $726,000 $564,000 $629,000 $739,000 $844,000 Contingency (30%) $1,189,000 $924,000 $1,030,000 $1,210,000 $1,382,000 General Conditions: Mobilization & $189,000 $147,000 $164,000 $193,000 $220,000 Demobilization (5%) Engineering, Administration, and Legal $928,000 $721,000 $804,000 $944,000 $1,078,000 (18%) Total Capital Costs ($) 6,081,000 4,721,000 5,266,000 6,187,000 7,066,000 Unit Capital Costs 6.88 5.05 5.72 6.64 7.68 ($/gpd) Total Power Cost ($/year) 84,000 28,000 20,000 44,000 28,000 Total Chemical Cost 69,000 67,000 67,000 67,000 67,000 ($/year) Total Labor Cost ($/year) 126,000 126,000 126,000 126,000 126,000 Total Replacement Cost 17,000 14,000 13,000 14,000 13,000 ($/year) Contingency (20%) 59,000 47,000 45,000 50,000 47,000 Total O&M Costs 353,000 280,000 269,000 299,000 278,000 Unit O&M Costs ($/kgal) 1.09 0.82 0.80 0.88 0.83 Total Life Cycle Costs ($) 12,800,000 10,100,000 10,400,000 11,900,000 12,400,000

The estimated treatment costs of the electrodialysis processes were proven lower than using UF/RO to treat reclaimed water for golf course irrigation. The economic evaluation based on the WATSYSTM data revealed a more than 21% total cost reduction while satisfying the product water quality requirements (shown in Tables2 and3). The total life cycle cost was reduced by approximately 19% using 2-stage EDR system based on the testing data. The chloride concentration was below the target goal of 150 mg/L. However, the sodium concentration cannot meet the operating goal of 110 mg/L, though very close to the contractual goal of 125 mg/L based on the pilot-scale testing results. Therefore a 4-stage EDR system was designed to meet both water quality requirements for sodium of 110 mg/L and chloride of 150 mg/L as well as to reduce the costs for treating reclaimed water for golf course irrigation. For a 4-stage EDR system, the unit O&M cost would be reduced to $0.83/kgal ($0.22/m3) from $1.09/kgal ($0.29/m3) in the baseline condition. The total life cycle cost was reduced to $11,900,000 and $12,400,000 in the WATSYSTM simulation and evaluation based on the testing data, respectively, in comparison to $12,800,000 in the UF/RO baseline. In addition to the reduced treatment costs, the EDR system has the advantage of achieving high water recovery: 90% unit water recovery and 92–93% overall water recovery considering water blending, while the UF/RO system could only achieve 85% unit water recovery and 88% overall water recovery. The TDS concentration of the product water of the EDR system is estimated better than the RO system: 433 mg/L for the 4-stage EDR system vs. 530 mg/L for UF/RO. In summary, the techno-economic modeling indicates that using electrodialysis for treating reclaimed water is more cost-effective and can recover more water than the RO process. Membranes 2021, 11, 333 17 of 20

4. Conclusions In this study, bench- and pilot-scale electrodialysis experiments were conducted to assess the technical feasibility and economic viability of reclaimed water desalination. The study elucidated the effects of various operating conditions, such as flow rate, HRT, applied current density, staging, and water recovery, on overall salt removal and monovalent permselectivity during reclaimed water treatment using electrodialysis. The overall negative impact of a higher flow rate was observed at the current density below 5.6 mA/cm2 for the bench-scale experiments. In comparison, the impact of a higher flow rate with increased linear velocity demonstrated a dominant positive impact at the initial current density tested in the pilot-scale electrodialysis. The linear trend of salt removal efficiency and salt flux was observed in both bench- and pilot-scale experiments, with overlapping linear trends of salt flux and scalable desalination efficiency (matching the ratio of residence time) between the bench- and pilot-scale electrodialysis systems. Such linear correlation of the salt flux and electrical charge density can be utilized to predict solute mass transport under different operating conditions with the same IEMs and spacers installed in the electrodialysis stack. The pilot-scale EDR system using the normal grade membranes demonstrated the technical feasibility of treating reclaimed water for golf course irrigation and surface discharge. An increase in the applied voltage enhanced the removal efficiency of ion separation and reduced the salt content in the product water. The target goal of sodium concentration 110 mg/L in the product water can be met at the current density of 2.36 mA/cm2 for the 1st electrical stage and 1.5 mA/cm2 for the 2nd electrical stage, while chloride concentration in product water was significantly below the requirement (<100 mg/L). Monovalent- 2+ 2+ 2− selectivity in terms of equivalent removal of divalent ions (Ca , Mg , and SO4 ) over monovalent ions (Na+, Cl−) was affected by both current density and water recovery. The salt rejection decreased in the pilot-scale EDR system at higher water recovery, due to a higher concentration gradient between the diluate and concentrate streams. No significant scaling/fouling on the IEM surface in the pilot-scale electrodialysis system was observed throughout the three-month continuous testing period. Total cost reduction can be achieved in electrodialysis, compared to the existing treatment option. The unit O&M cost would be reduced to $0.83/kgal ($0.22/m3) from $1.09/kgal ($0.29/m3) in the baseline condition, along with reduced total life cycle cost. The water recovery of the EDR system is higher than the UF/RO, with 90% unit water recovery and 92–93% overall water recovery. In conclusion, the electrodialysis demonstrated promising technical feasibility and economic viability for treating reclaimed water.

Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/membranes11050333/s1. Figure S1. Desalination performance as a function of current density in terms of ion concentration, ion removal, conductivity, conductivity reduction, and pH change. Reference [58] is cited in supplementary material. Author Contributions: Conceptualization, X.X., Q.H., G.M., H.W., N.N. and P.X.; methodology, X.X., Q.H., G.M., H.W., N.N. and P.X.; data analysis and visualizations, X.X., Q.H., G.M., H.W. and P.X.; writing—original draft preparation, X.X., Q.H.; writing—review and editing, X.X. and P.X.; supervision, H.W. and P.X.; project administration, N.N. and P.X.; funding acquisition, N.N. and P.X. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by the U.S. Department of the Interior, Bureau of Reclama- tion (R14AP00175), and the National Science Foundation Engineering Research Center ReNUWIt (EEC-1028968). Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: Detailed experimental data can be provided upon reasonable request. Membranes 2021, 11, 333 18 of 20

Acknowledgments: The authors thank Patrick Girvin, Danny Jorgensen, Yongchang Zheng, John Barber, and Ken Irwin with Suez Water Technologies & Solutions for technical support of the electrodialysis systems; and Art Nunez, Laura McCasland, and Suzanne Grendahl for supporting the pilot-scale testing and assistance in chemical analysis in the Scottsdale Water Campus, AZ. Conﬂicts of Interest: The authors declare no conﬂict of interest.

References 1. Greve, P.; Kahil, T.; Mochizuki, J.; Schinko, T.; Satoh, Y.; Burek, P.; Fischer, G.; Tramberend, S.; Burtscher, R.; Langan, S.; et al. Global assessment of water challenges under uncertainty in water scarcity projections. Nat. Sustain. 2018, 1, 486–494. [CrossRef] 2. Nobre, R.L.G.; Caliman, A.; Cabral, C.R.; de Carvalho Araújo, F.; Guerin, J.; Dantas, F.D.C.C.; Quesado, L.B.; Venticinque, E.M.; Guariento, R.D.; Amado, A.M.; et al. Precipitation, landscape properties and land use interactively affect water quality of tropical freshwaters. Sci. Total Environ. 2020, 716, 137044. [CrossRef] 3. MacDonald, G.M. Water, climate change, and sustainability in the southwest. Proc. Natl. Acad. Sci. USA 2010, 107, 21256–21262. [CrossRef] 4. Pettygrove, G.S. Irrigation with Reclaimed Municipal Wastewater—A Guidance Manual; CRC Press: Boca Raton, FL, USA, 2018. 5. Zhang, Y.; Shen, Y. Wastewater irrigation: Past, present, and future. Wiley Interdiscip. Rev. Water 2019, 6, e1234. [CrossRef] 6. Fournier, E.D.; Keller, A.A.; Geyer, R.; Frew, J. Investigating the energy-water usage efficiency of the reuse of treated municipal wastewater for artificial groundwater recharge. Environ. Sci. Technol. 2016, 50, 2044–2053. [CrossRef][PubMed] 7. Zuurbier, K.; Smeets, P.; Roest, K.; van Vierssen, W. Use of Wastewater in Managed Aquifer Recharge for Agricultural and Drinking Purposes: The Dutch Experience. In Safe Use of Wastewater in Agriculture; Springer: New York, NY, USA, 2018; pp. 159–175. 8. Wei, X.; Binger, Z.M.; Achilli, A.; Sanders, K.T.; Childress, A.E. A modeling framework to evaluate blending of seawater and treated wastewater streams for synergistic desalination and potable reuse. Water Res. 2020, 170, 115282. [CrossRef][PubMed] 9. Walker, M.E.; Theregowda, R.B.; Safari, I.; Abbasian, J.; Arastoopour, H.; Dzombak, D.A.; Hsieh, M.K.; Miller, D.C. Utilization of municipal wastewater for cooling in thermoelectric power plants: Evaluation of the combined cost of makeup water treatment and increased condenser fouling. Energy 2013, 60, 139–147. [CrossRef] 10. Council, N.R. Water Reuse: Potential for Expanding the Nation’s Water Supply through Reuse of Municipal Wastewater; National Academies Press: Washington, DC, USA, 2012. 11. Evanylo, G.; Ervin, E.; Zhang, X. Reclaimed water for turfgrass irrigation. Water 2010, 2, 685–701. [CrossRef] 12. Phogat, V.; Mallants, D.; Cox, J.W.; Šim ˚unek,J.; Oliver, D.P.; Pitt, T.; Petrie, P.R. Impact of long-term recycled water irrigation on crop yield and soil chemical properties. Agric. Water Manag. 2020, 237, 106167. [CrossRef] 13. Qian, Y.; Mecham, B. Long-term effects of recycled wastewater irrigation on soil chemical properties on golf course fairways. Agron. J. 2005, 97, 717–721. [CrossRef] 14. Suarez, D.L.; Wood, J.D.; Lesch, S.M. Infiltration into cropped soils: Effect of rain and sodium adsorption ratio–impacted irrigation water. J. Environ. Qual. 2008, 37 (Suppl. S5), S-169–S-179. [CrossRef] 15. Goodfellow, W.L.; Ausley, L.W.; Burton, D.T.; Denton, D.L.; Dorn, P.B.; Grothe, D.R.; Heber, M.A.; Norberg-King, T.J.; Rodgers, J.H., Jr. Major ion toxicity in effluents: A review with permitting recommendations. Environ. Toxicol. Chem. Int. J. 2000, 19, 175–182. [CrossRef] 16. Pype, M.-L.; Lawrence, M.G.; Keller, J.; Gernjak, W. Reverse osmosis integrity monitoring in water reuse: The challenge to verify virus removal—A review. Water Res. 2016, 98, 384–395. [CrossRef] 17. Qasim, M.; Badrelzaman, M.; Darwish, N.N.; Darwish, N.A.; Hilal, N. Reverse osmosis desalination: A state-of-the-art review. Desalination 2019, 459, 59–104. [CrossRef] 18. Xu, P.; Cath, T.Y.; Robertson, A.P.; Reinhard, M.; Leckie, J.O.; Drewes, J.E. Critical Review of Desalination Concentrate Manage- ment, Treatment and Beneficial Use. Environ. Eng. Sci. 2013, 30, 502–514. [CrossRef] 19. Llanos, J.; Cotillas, S.; Cañizares, P.; Rodrigo, M.A. Novel electrodialysis–electrochlorination integrated process for the reclamation of treated wastewaters. Sep. Purif. Technol. 2014, 132, 362–369. [CrossRef] 20. Xu, X.; Lin, L.; Ma, G.; Wang, H.; Jiang, W.; He, Q.; Nirmalakhandan, N.; Xu, P. Study of Polyethyleneimine Coating on Membrane Permselectivity and Desalination Performance during Pilot-scale Electrodialysis of Reverse Osmosis Concentrate. Sep. Purif. Technol. 2018, 207, 396–405. 21. Kwon, K.; Han, J.; Park, B.H.; Shin, Y.; Kim, D. Brine recovery using reverse electrodialysis in membrane-based desalination processes. Desalination 2015, 362, 1–10. [CrossRef] 22. American Water Works Association. Electrodialysis and Electrodialysis Reversal (M38); American Water Works Association: Denver, CO, USA, 1995; Volume 38. 23. Hansima, M.; Makehelwala, M.; Jinadasa, K.; Wei, Y.; Nanayakkara, K.; Herath, A.; Weerasooriya, R. Fouling of ion exchange membranes used in the electrodialysis reversal advanced water treatment: A review. Chemosphere 2020, 127951. 24. McGovern, R.K.; Weiner, A.M.; Sun, L.; Chambers, C.G.; Zubair, S.M. On the cost of electrodialysis for the desalination of high salinity feeds. Appl. Energy. 2014, 136, 649–661. [CrossRef] 25. McGovern, R.K.; Zubair, S.M.; Lienhard, V.J.H. The cost effectiveness of electrodialysis for diverse salinity applications. Desalina- tion 2014, 348, 57–65. [CrossRef] Membranes 2021, 11, 333 19 of 20

26. Xu, P.; Capito, M.; Cath, T.Y. Selective removal of arsenic and monovalent ions from brackish water reverse osmosis concentrate. J. Hazard. Mater. 2013, 260, 885–891. [CrossRef][PubMed] 27. Geraldes, V.; Afonso, M.D. Limiting current density in the electrodialysis of multi-ionic solutions. J. Membr. Sci. 2010, 360, 499–508. [CrossRef] 28. Sadrzadeh, M.; Razmi, A.; Mohammadi, T. Separation of different ions from wastewater at various operating conditions using electrodialysis. Sep. Purif. Technol. 2007, 54, 147–156. [CrossRef] 29. Zhang, Y.; Desmidt, E.; Van Looveren, A.; Pinoy, L.; Meesschaert, B.; Van der Bruggen, B. Phosphate separation and recovery from wastewater by novel electrodialysis. Environ. Sci. Technol. 2013, 47, 5888–5895. [CrossRef] 30. Abou-Shady, A. Recycling of polluted wastewater for agriculture purpose using electrodialysis: Perspective for large scale application. Chem. Eng. J. 2017, 323, 1–18. [CrossRef] 31. Arola, K.; Ward, A.; Mänttäri, M.; Kallioinen, M.; Batstone, D. Transport of pharmaceuticals during electrodialysis treatment of wastewater. Water Res. 2019, 161, 496–504. [CrossRef] 32. Mohammadi, R.; Ramasamy, D.L.; Sillanpää, M. Enhancement of nitrate removal and recovery from municipal wastewater through single-and multi-batch electrodialysis: Process optimisation and energy consumption. Desalination 2020, 498, 114726. [CrossRef] 33. Al-Amshawee, S.; Yunus, M.Y.B.M.; Azoddein, A.A.M.; Hassell, D.G.; Dakhil, I.H.; Hasan, H.A. Electrodialysis desalination for water and wastewater: A review. Chem. Eng. J. 2020, 380, 122231. [CrossRef] 34. Gurreri, L.; Tamburini, A.; Cipollina, A.; Micale, G. Electrodialysis applications in wastewater treatment for environmental protection and resources recovery: A systematic review on progress and perspectives. Membr 2020, 10, 146. [CrossRef] 35. Walker, W.S.; Kim, Y.; Lawler, D.F. Treatment of model inland brackish groundwater reverse osmosis concentrate with electrodialysis—Part II: Sensitivity to voltage application and membranes. Desalination 2014, 345, 128–135. [CrossRef] 36. Lee, H.J.; Sarfert, F.; Strathmann, H.; Moon, S.H. Designing of an electrodialysis desalination plant. Desalination 2002, 142, 267–286. [CrossRef] 37. Walker, W.S.; Kim, Y.; Lawler, D.F. Treatment of model inland brackish groundwater reverse osmosis concentrate with electrodialysis—Part I: Sensitivity to superficial velocity. Desalination 2014, 344, 152–162. [CrossRef] 38. Warsinger, D.M.; Swaminathan, J.; Guillen-Burrieza, E.; Arafat, H.A. Scaling and fouling in membrane distillation for desalination applications: A review. Desalination 2015, 356, 294–313. [CrossRef] 39. Walker, W.S.; Kim, Y.; Lawler, D.F. Treatment of model inland brackish groundwater reverse osmosis concentrate with electrodialysis—Part III: Sensitivity to composition and hydraulic recovery. Desalination 2014, 347, 158–164. [CrossRef] 40. Myint, M.T. Saving energy consumption and CO2 emission from sustainable efficient operating zones in inland electrodialysis reversal desalination. Water Resour. Ind. 2014, 5, 36–48. [CrossRef] 41. Krol, J.; Wessling, M.; Strathmann, H. Concentration polarization with monopolar ion exchange membranes: Current–voltage curves and water dissociation. J. Membr. Sci. 1999, 162, 145–154. [CrossRef] 42. Rubinstein, I.; Warshawsky, A.; Schechtman, L.; Kedem, O. Elimination of acid-base generation (‘water-splitting’) in electrodialysis. Desalination 1984, 51, 55–60. [CrossRef] 43. Lee, H.-J.; Strathmann, H.; Moon, S.-H. Determination of the limiting current density in electrodialysis desalination as an empirical function of linear velocity. Desalination 2006, 190, 43–50. [CrossRef] 44. Cowan, D.A.; Brown, J.H. Effect of turbulence on limiting current in electrodialysis cells. Ind. Eng. Chem. 1959, 51, 1445–1448. [CrossRef] 45. Tanaka, Y. Current density distribution and limiting current density in ion-exchange membrane electrodialysis. J. Membr. Sci. 2000, 173, 179–190. [CrossRef] 46. Xu, X.; He, Q.; Ma, G.; Wang, H.; Nirmalakhandan, N.; Xu, P. Selective separation of mono-and di-valent cations in electrodialysis during brackish water desalination: Bench and pilot-scale studies. Desalination 2018, 428, 146–160. [CrossRef] 47. Todeschini, S.; Perreault, V.; Goulet, C.; Bouchard, M.; Dubé, P.; Boutin, Y.; Bazinet, L. Assessment of the Performance of Electrodialysis in the Removal of the Most Potent Odor-Active Compounds of Herring Milt Hydrolysate: Focus on Ion-Exchange Membrane Fouling and Water Dissociation as Limiting Process Conditions. Membranes 2020, 10, 127. [CrossRef] 48. Tanaka, Y. Concentration polarization in ion-exchange membrane electrodialysis—The events arising in a flowing solution in a desalting cell. J. Membr. Sci. 2003, 216, 149–164. [CrossRef] 49. Sadrzadeh, M.; Razmi, A.; Mohammadi, T. Separation of monovalent, divalent and trivalent ions from wastewater at various operating conditions using electrodialysis. Desalination 2007, 205, 53–61. [CrossRef] 50. Lee, H.-J.; Song, J.-H.; Moon, S.-H. Comparison of electrodialysis reversal (EDR) and electrodeionization reversal (EDIR) for water softening. Desalination 2013, 314, 43–49. [CrossRef] 51. Han, L.; Galier, S.; Roux-de Balmann, H. Ion hydration number and electro-osmosis during electrodialysis of mixed salt solution. Desalination 2015, 373, 38–46. [CrossRef] 52. Ma, G.; Xu, X.; Tesfai, M.; Wang, H.; Xu, P. Developing anti-biofouling and energy-efficient cation-exchange membranes using conductive polymers and nanomaterials. J. Membr. Sci. 2020, 603, 118034. [CrossRef] 53. Cooke, B. Concentration polarization in electrodialysis—I. The electrometric measurement of interfacial concentration. Electrochim. Acta. 1961, 3, 307–317. [CrossRef] Membranes 2021, 11, 333 20 of 20

54. Valero, F.; Arbós, R. Desalination of brackish river water using Electrodialysis Reversal (EDR): Control of the THMs formation in the Barcelona (NE Spain) area. Desalination 2010, 253, 170–174. [CrossRef] 55. Sata, T.; Teshima, K.; Yamaguchi, T. Permselectivity between two anions in anion exchange membranes crosslinked with various diamines in electrodialysis. J. Polym. Sci. Part A Polym. Chem. 1996, 34, 1475–1482. [CrossRef] 56. Zhang, Y.; Ghyselbrecht, K.; Meesschaert, B.; Pinoy, L.; Van der Bruggen, B. Electrodialysis on RO concentrate to improve water recovery in wastewater reclamation. J. Membr. Sci. 2011, 378, 101–110. [CrossRef] 57. Kwon, B.; Lee, S.; Cho, J.; Ahn, H.; Lee, D.; Shin, H.S. Biodegradability, DBP formation, and membrane fouling potential of natural organic matter: Characterization and controllability. Environ. Sci. Technol. 2005, 39, 732–739. [CrossRef][PubMed] 58. Christensen, P.; Dysert, L.R.; Bates, J.; Burton, D.; Creese, R.C.; Hollmann, J. Cost Estimate Classiﬁcation System—As Applied in Engineering, Procurement, and Construction for the Process Industries; AACE, Inc: Morgantown, WV, USA, 2011.