This study evaluated the feasibility of forward osmosis (FO) in diluting and reusing the concentrate produced in a reverse osmosis (RO) plant in James City County, VA. Secondary treated wastewater (STW) was used as the feed solution. Findings indicated that pH had slight effects on the water flux of the FO membrane. As the concentration of total dissolved solids (TDS) in the concentrate was diluted from 12.5 to 1.0 g/L or the temperature in the STW decreased from 23 to 10 °C, the membrane flux decreased from 2.2 to 0.59 and 0.81 L/(m2 h), respectively. The FO membrane showed a good performance in the rejection of organic pollutants, with only a small part of the protein-like substances and disinfection byproducts permeating to the diluted concentrate. During an 89-hour continuous operation, water flux decline due to membrane fouling was not observed. Controlling the TDS in the second-stage FO effluent at 1.5 g/L, approximately 8.3% of the pump energy input could be saved. The consumption of groundwater was reduced from 22.7 × 103 to 10.6 × 103 m3/d. FO was proved to be an effective method in both diluting the discharged concentrate and reducing the energy consumption of RO.
With the increase in water demand, many utilities are turning to non-traditional water, such as brackish water, seawater, and even wastewater, as a possible water source in many countries and areas of the world (National Research Council (US) 2008). Accordingly, reverse osmosis (RO) is becoming more and more common as a technique to produce drinking water because of its high efficiency in pollutant removal (Herzberg et al. 2009). This is the case for James City County in Virginia, USA, where groundwater with a high salt content (brackish water) is the main water source. The raw water is desalinated by a RO system. One of the major issues that the actual process presents is disposal of the concentrate. Due to the environmental impacts of high salinity water, the discharge of desalination concentrated solutions is highly controlled by the regulatory agencies. Strict regulations have been created over the last decade, making it more difficult to build a new facility or retrofit/expand existing ones.
In the RO process, hydraulic pressure is used to oppose and exceed the osmotic pressure of a saline aqueous feed solution. The applied pressure is the driving force for the mass transport through a semi-permeable membrane (Greenlee et al. 2009; Malaeb & Ayoub 2011). The membrane allows the passage of clean water, while the salt and other contaminants are held back (National Research Council (US) 2008). Accordingly, both the energy consumed in the RO unit and the salt content in the discharged concentrate are high (Stoughton & Lietzke 1965; Lattemann & Höpner 2008; Semiat 2008; Elimelech & Phillip 2011). Although great gains have been made over the last decade in the RO technology to increase its energy efficiency, energy costs still contribute as much as 75% of the operation cost of desalination plants (Semiat 2008). Innovations that help to reduce energy consumption and the amount of concentrate discharged will strengthen the suitability of RO in drinking water purification.
The amount of energy used in RO mainly depends on the salt concentration of the feed solution, and can range from 3.0 to 7.0 kWh/m3 for seawater desalination (National Research Council (US) 2008). For brackish water desalination, however, where the concentration of salt (1–10 g/L) is notably lower than that of the seawater (10–30 g/L), the energy consumption is approximately 0.5–3.0 kWh/m3 (National Research Council (US) 2008). For a feed solution with low salt content, a notable decrease in the osmotic pressure that must be overcome with applied hydraulic pressure can be achieved (Semiat 2008; Subramani et al. 2011). Accordingly, the RO process would require less energy to desalinate diluted brackish water than it does for the raw brackish water, which is the upgrading plan proposed by the authors for the upgrading of the existing RO system in James City County. The secondary treated wastewater (STW) was used as the dilution water.
Compared with direct dilution, forward osmosis (FO) can enable the use of impaired water to dilute the water entering the RO plant with high pollutant rejection efficiency (Cath et al. 2006, 2010). In the FO unit, a low salt concentration water source, acting as the feed solution, is separated by a selective FO membrane from water with a high salt concentration, which acts as the draw solution. The two solutions are placed on opposite sides of the membrane. Fresh water will move from the feed solution towards the draw solution, leaving contaminants retained in the membrane (McCutcheon & Elimelech 2007; Qin et al. 2010). Accordingly, FO is probably a competitive technique in diluting the influent of the RO system.
However, the quality of the STW differs from traditional water sources and has limited usages because of the presence of toxic organic materials (Carrara et al. 2008), which pose a proven or potentially high health risk to humans. Previous studies have mainly focused on the rejection performance of FO for a few parts of specific hazardous materials, such as pharmaceuticals, personal care products, and flame-retardants (Hancock et al. 2011; Kong et al. 2014, 2015). The application of STW as an indirect source water in the hybrid FO/RO system and the organic rejection performance of FO have not yet been studied in depth. The objective of this work, therefore, is to investigate (i) the water flux variation law with the dilution of the concentrate and the variation in temperature and pH of the STW, (ii) the organic rejection performance of the FO membrane, and (iii) the feasibility of FO in diluting and reusing the concentrate produced in the RO unit using STW as the source water.
Existing RO desalination process in James City County, VA
Four RO skids are installed in the treatment facility for the desalination of the raw water extracted from the Lower Potomac Wells. The RO skids consist of a two-stage membrane system. The first stage consists of 20 vessels, and the second stage consists of 10 additional vessels. The vessels contain six membranes each. Groundwater is pumped into the first stage membranes of the RO skid. The permeate is discharged to the permeate line. The concentrate from the first stage RO becomes the feed water for the second stage RO. The permeate from the second stage combines with the permeate from the first stage. The concentrate from the second stage RO is piped to the concentrate line.
The raw water used in the drinking water plant is pumped from five wells. One of them supplies approximately 3.8 × 103 m3/d groundwater with low salt concentration (Middle Potomac Well). The other four wells supply approximately 18.9 × 103 m3/d groundwater with a higher salt concentration (Lower Potomac Well). The water quality of both types of wells is presented in Table 1. Treatment with RO removes nearly all of the salts in the Lower Potomac Well water. To maintain a normal range of salt contents in the potable water, the RO effluent is blended with the groundwater pumped from the Middle Potomac Well as shown in Figure 1. The concentration of total dissolved solids (TDS) in the brackish water is approximately 2,500 mg/L, which increases to approximately 13,000 mg/L after being concentrated. To avoid the discharge of this highly concentrated brackish water, a two-stage FO is suggested for the upgrading of the RO system in James City County.
|Water quality parameters||Middle Potomac Well||Lower Potomac Well||Concentrate||STW|
|Water quality parameters||Middle Potomac Well||Lower Potomac Well||Concentrate||STW|
–, data not available; ND, under determination limit.
EXPERIMENTAL MATERIALS AND METHODS
Feed and draw solution of the FO system
The concentrate, which was used as the draw solution, was obtained from the RO system in James City County, VA. The STW, provided by a wastewater treatment plant in Middletown, PA, was used as the feed solution, as the plant is conveniently located near the Harrisburg campus where the experiments were conducted. This wastewater treatment plant consists of a solids grinder and grit remover, primary clarification, biological treatment with clarification, chlorine addition for disinfection, and de-chlorination prior to discharge. The quality of the concentrate and the STW are listed in Table 1.
FO cross flow setup
Feed and draw solutions were continuously circulated between the storage tanks and the membrane cell. The initial volumes of the feed and draw solutions were 350 mL and 1,000 mL, respectively. The cross-flow rate on each side of the membrane was the same, and was controlled at 45 mL/min with a variable speed pump (Cole-Parmer, Vernon Hills, IL). Changes in the weight of the draw solution were recorded by a digital balance (TL2100, Mettler Toledo, Germany) to determine the membrane flux of the pure water. The active layer of the FO membrane was facing the feed solution throughout the experiment.
To examine the effect of pH on the flux of water, 0.1 mol/L NaOH was added to the feed solution to increase its pH from 7.5 to 8.5. The total running time was controlled at approximately 15 min. To determine the effects of the water temperature on the flux of the FO membrane, the temperature of the feed solution was adjusted between 3 and 30 °C in a constant-temperature incubator. To assess the effects of membrane fouling on water flux, the variation of the water flux of the FO membrane was monitored in 89 h of continuous running. The pH and temperature of the draw solution were maintained at 8.0 and 23 °C, respectively.
Water quality and membrane structure analysis methods
The trihalomethanes (THMs) and haloacetic acids (HAAs) were analyzed by gas chromatography with an electron capture detector (Agilent 7890A, Santa Clara, USA) after liquid-liquid extraction and methylation (for HAAs only) according to the standard method of EPA551.1 and EPA552.3, respectively. Standard solutions of four THMs, including CHCl3, bromodichlorobromomethane (CHCl2Br), chlorodibromomethane (CHBr2Cl), and bromoform (CHBr3), and six HAAs, including monochloroacetic acid, monobromoacetic acid, dichloroacetic acid (DCAA), bromochloroacetic acid (BCAA), dibromoacetic acid, and trichloroacetic acid (TCAA), were purchased from Sigma-Aldrich (Germany). pH and temperature were monitored on-line using a Logger Pro 3.8.6 with corresponding sensors. Before measuring, all sensors were calibrated with standard solutions. TDS was measured according to standard method 2540c (APHA et al. 2005). The concentration of organic matters was measured by a total organic carbon (TOC) analyzer (TOC-VCPH, Shimadzu, Japan) and a spectrometer (T6, Puxi, China) at 254 nm.
Fluorescence measurements were conducted using a spectrofluorometer (FP-6500, Jasco, Japan) equipped with a 150 W xenon lamp at 23 °C. A 1.0 cm quartz cuvette with four optical windows was used for the analyses. An emission scan was conducted from 250 to 600 nm at a wavelength step of 5 nm and excitation wavelengths from 220 to 450 nm at a 5 nm interval. The detector was set to high sensitivity, and the scanning speed was maintained at 2,000 nm/min. To observe the surface characteristics of the FO membrane, scanning electron microscopy (SEM) (JSM-6490LV, JEOL Ltd, Japan) was applied. The SEM samples were sputter-coated with gold before conducting scanning.
RESULTS AND DISCUSSION
Effects of TDS in the concentrate on the water flux of the FO membrane
Effects of pH and temperature in the STW on the water flux of the FO membrane
Compared with pH, temperature showed a more significant effect on the water flux through the FO membrane. As shown in Figure 4(b), the average water flux was approximately 2.75 L/(m2·h) when the temperature of the draw solution was 30 °C, corresponding to the STW temperature in summer, and decreased notably to approximately 0.81 L/(m2·h) at 10 °C, corresponding to the STW temperature in winter. A linear relationship was found between the membrane flux and the solution temperature. The slope and intercept of the best fit line were calculated to be 0.102 and –0.372 respectively, which was close to the values (0.064 and –0.227) obtained by Wang et al. (2014) using rainwater and cooling water as the feed and draw solutions, respectively.
However, based on the van't Hoff equation (You et al. 2012), the osmotic pressure of the feed solution decreases with the decrease of its temperature, which will increase the net osmotic pressure for driving the water in the FO process. Besides net osmotic pressure, both the absolute and kinematic viscosities of the feed solution increased notably with the decrease of feed solution temperature (Phuntsho et al. 2012). Kim et al. (2015) determined that as the solution temperature decreased from 50 °C to 20 °C, osmotic coefficients and diffusivity decreased by 6.6% and 48.3% respectively, which increased the internal concentration polarization (ICP). The increased ICP and solution viscosity was probably the major reason that inhibited the permeation of pure water at low temperatures.
Organic rejection performance of the FO membrane
Besides UV254 and TOC, a 3D excitation–emission matrix fluorescence spectrum was used to investigate the type of organic matter that passed through the FO membrane. The organic matter contained in the STW (feed solution) was rich in humic acid-like substances (Figure 5(b)), associated with the peak of Ex/Em = 420–440/235–245 nm in the fluorescence spectrum (Henderson et al. 2009). Compared with the STW, the concentrate (draw solution) was low in organic pollutants, and no peak associated with organic matter was observed (Figure 5(c)). After 8 h of operation, only one peak (Ex/Em = 375/350 nm) contributed by protein-like substances (Reynolds 2002; Liu et al. 2011) relating to the activity of bacteria was observed. This further proved that the FO membrane used in this study had a high capacity to remove organic matter (Figure 5(d)).
The removal rates for THMs were above 95%, higher than that of HAAs (50–60%) under the selected pH conditions (Figure 6(a)). The low rejection rate of HAAs was probably related to its low initial concentration in the STW used in our study (Xie 2003). These rejections are comparable to the values reported for NF membranes (56–100%) and RO membranes (86–94% for HAAs) (Agus & Sedlak 2010). We speculate that the mechanisms underlying the removal of THMs and HAAs might be size exclusion and charge repulsion, similar to the NF/RO membranes (Bellona et al. 2004; Kong et al. 2014). Unlike HAAs, all of the THMs were highly volatile organic materials (Lee et al. 2009). As the hydraulic retention time of the STW in the feed solution tank increased, part of the THMs might enter the air under the effect of volatilization. Further study showed that the concentration of THMs mainly existing in the forms of CHCl2Br and CHCl3 decreased notably from 101 to 6.7 μg/L after 4 days of retention (without stirring) in the feed solution tank (Figure 6(b)). Compared with THMs, the concentration of HAAs, mainly existing in the forms of DCAA, TCAA, and BCAA, was relatively stable (Figure 6(c)).
Feasibility analysis of FO in brackish water desalination system upgrading
Compared with RO, the water flux of FO was relatively small, especially for the draw solution with low salt content. To guarantee the average water flux, a two-stage FO in which the TDS in the concentrate was diluted from 12.5 to 1.5 g/L in two series-connection draw solution tanks is suggested for the upgrading of the existing RO system in James City County, as shown in Figure 1. STW was used as a feed solution. By controlling the retention time of the concentrate in the draw solution tank, the TDS in the first-stage FO effluent was maintained at 6.0 g/L, corresponding to an increase in the volume from 3.8 × 103 to 7.6 × 103 m3/d. Sixty per cent of the first-stage FO effluent was blended with 9.2 × 103 m3/d of the STW directly before it was discharged to the James River. In the second-stage FO, 40% of the first-stage FO effluent was further diluted four times. The effluent, with 1.5 g/L of TDS, was reused as the influent of the RO unit. The consumption of the groundwater could be reduced from 22.7 × 103 to 10.6 × 103 m3/d.
Considering the concentration of organic matter permeated into the diluted concentrate, part of which was reused as the influent for the RO, was limited based on accumulative values in 8 h, the effects of organic matter in the STW on the quality of the treated drinking water could be ignored. Before upgrading, the pump pressure needed in the RO unit was approximately 0.6 MPa. Based on the water quality shown in Table 1, NaCl and NaHCO3 were assumed to be contributing to the osmotic pressure generated by the brackish water. Calculation results showed that after upgrading the osmotic pressure existing between the two sides of the RO membrane would decrease from 0.18 to 0.13 MPa so as the salt content in the influent decreased from 2.3 to 1.78 g/L. Maintaining a water purification capacity of 22.7 × 103 m3/d, the reduced pump pressure was determined to be 0.05 MPa, so approximately 8.3% of the input energy of the pump station could be saved.
Over the last decades, strict regulations have been established to protect water bodies by limiting the amount of TDS that are discharged into waterways. In this study, the implementation of FO was considered to treat and reuse the concentrate from the RO process. Solution pH showed a slight effect on the flux of both pure water and DBPs. With the decrease in water temperature, especially in winter, the water flux of the FO membrane decreased notably. Although part of the protein-like substances permeated into the draw solution, the FO membrane used in this study showed a good performance in the rejection of organic pollutants. More than 98% of TOC, 95% of THMs, and 50–60% of HAAs were removed from the feed solution. Considering that these removals were based on accumulative values in 8 h, the detrimental effect of using STW as source water on the quality of the produced drinking water could be ignored. Furthermore, during an 89-hour operation, there was no observed flux decline due to membrane fouling. By controlling the effluent concentration of TDS in the second stage of FO at 1.5 g/L, the consumption of groundwater would decrease to 10.6 × 103 m3/d. Approximately 8.3% of the pump energy input was saved. Meanwhile, the environmental impact caused by the discharge of concentrate into the James River could be minimized.
This research was supported by the National Natural Science Foundation of China (No. 21007050), the Science and Technology Nova Program of Shaanxi (No. 2014KJXX-66), and the open funds of Zhejiang Provincial Key Laboratory of Water Science and Technology.