Abstract

The low cost simplified method for implementation of pressure-assisted osmotic (PAO) backwash (BW) for spiral wound reverse osmosis (RO) membrane module is presented in this work. The effect of membrane design and an operating parameter concerning the efficiency of PAO membrane BW is analyzed. The following design and operating parameters are considered in this study: (i) spacer thickness, (ii) dimension of the permeate channel, (iii) number of leaves, and (iv) BW water pressure. The performance of PAO BW with respect to membrane cleaning efficiency is analyzed for three different high recovery RO systems by purifying 1,500 liters of water. The membrane cleaning efficiency is measured by examining the rate of permeate quality and quantity decline using ASTM D4516 method. Finally, to quantify the membrane fouling with respect to different high recovery configurations, the thickness, and composition of foulants present in the used membrane's surface are measured by using field emission scanning electron microscope with energy dispersive x-ray (FESEM-EDX). The result concludes that the RO membrane operated at high recovery with PAO BW is found to have less fouling deposits than membrane without PAO BW.

INTRODUCTION

Reverse osmosis (RO) is one of the widely used techniques for water purification and seawater desalination (Gao et al. 2016). RO produces high-quality water by removing micro-pollutants, organic-inorganic matter colloids, salts, and pathogens (Salvador Cob et al. 2012). The recovery of the RO process is defined as the ratio between feed and product water flow rate. Different configurations of the RO process are used in industry to achieve high recovery and process efficiency, such as a single stage, multistage, batch, and semi-batch. Although high recovery RO configuration is proven to be one of the minimum energy consuming desalination technologies, the process efficiency is hindered due to the concentration polarization (CP) and reuse of concentrated reject that leads to membrane fouling and scaling issues. The fouling/scaling causes membrane degradation, productivity loss, and results in high energy consumption (Spiegler & Macleish 1981; Ramon et al. 2010).

Furthermore, both batch and semi-batch RO configuration have been explored by many researchers in order to achieve high recovery in the single-stage RO configuration (Efraty 2012a, 2012b, 2015a, 2015b, 2015c, 2016a, 2016b, 2016c, 2016d; Efraty et al. 2012; Gal & Efraty 2016; Septon & Efraty 2016; Sonera et al. 2016). Simultaneously, the booster pump is used to minimize overall energy consumption. The RO batch configuration with brine recirculation and energy recovery is also named as a closed-circuit RO system (CCRO). In terms of the energy, both CCRO and batch RO were proven to work more efficiently than the multistage RO configuration without inter-stage booster pumps (Qiu & Davies 2012; Stover 2013; Warsinger et al. 2016). Stover (2013) proposed a single-stage CCRO with increased recovery up to 88%, whereas Sonera et al. (2016) achieved 93.8% recovery through a single stage, three element CCRO for brackish water desalination. In CCRO, the reduced membrane recovery and reject recycle concept can reduce bio-fouling, scaling of inorganic elements, and specific energy consumption (SEC) (Stover 2013; Sonera et al. 2016). However, due to the recirculation, the foulant concentration will increase with time. Once the foulant concentration reaches its saturation point, it starts to deposit on the membrane surface. As the feed concentration to the membrane increases, the water permeation across the membrane decreases. Thus excess feed pressure is applied to achieve the constant product flow rate, which leads to high operational cost and shorter lifespan of the membrane (Astudillo et al. 2010).

Depending upon the nature of the foulant, it can be decided whether it is reversible fouling or not. In general, reversible fouling is mitigated by membrane cleaning; however, long-term exposure to irreversible fouling leads to membrane replacement (Kim et al. 2007). Membrane fouling remains a significant problem in the RO process. Methods such as physical, chemical, and physicochemical cleaning are recommended to regenerate the membrane performance (Sagiv & Semiat 2005; Kim et al. 2007; Qin et al. 2010; Luján-Facundo et al. 2013). Physical cleaning methods such as forward or reverse flushing (Uchymiak et al. 2009), air sparing, backwashing, CO2 back permeation, ultrasonic (Lu et al. 2012; Mizrahi et al. 2012), magnetic field and an electric field can remove the reversible RO fouling components (Trägårdh 1989; Qin et al. 2010; Peiris et al. 2013). Chemical cleaning methods attribute to weaken cohesion forces between the membrane surface and foulants to remove irreversible fouling components (Sagiv & Semiat 2005; Porcelli & Judd 2010). In this type of cleaning, the selection of the cleaning agent is very crucial. Depending on the membrane material and kind of foulant, the choice of cleaning agent is decided. If both physical and chemical cleaning techniques are not sufficient, then the membrane has to be replaced to achieve the desired permeate water flow and concentration.

As reported in the literature, osmotic BW is found to be more useful to remove reversible fouling and reduces irreversible fouling potential, which could minimize the discharge of cleaning chemicals to the environment and their impact on the RO membrane (Ramon et al. 2010, 2013; Motsa et al. 2017). The membrane BW process will enable the permeate back-flow from the permeate side to feed side (Spiegler & Macleish 1981; Sagiv & Semiat 2005, 2010a, 2010b; Liberman 2009, 2017; Qin et al. 2009, 2010). For example, Sagiv & Semiat have performed an RO experiment with aqueous CaCO3 as a feed solution and explained the effect of the osmotic BW process on CaCO3 removal from the membrane surface. The concentration polarization layer is diluted when the BW water enters from the permeate channel to the feed channel, which favors cleaning of the membrane surface and recovers its original flowrate (Trägårdh 1989; Sagiv et al. 2008; Sagiv & Semiat 2010a, 2010b). The BW flow can be enabled either by conventional forward osmosis (FO) or by PAO. In the FO process, the driving force is the concentration difference between the feed and permeate channel. In the PAO process, both concentration and pressure difference across the membrane act as a driving force. In RO, PAO BW may lead to a membrane support layer peeling off at very high BW pressure. Therefore, the calculation of the minimal required BW pressure for PAO-based membrane cleaning is a crucial aspect. The dimensions of the permeate channel, such as permeate channel length, width, and spacer thickness are essential parameters to achieve uniform BW in the spiral-wound module, and those parameters have to be optimized. Permeate channel spacer thickness and the number of leaves in the spiral-wound module decide the uniformity of the BW flux. For example, a single leaf spiral-wound module is expected to provide less cleaning efficiency compared to the dual leaf module with equal surface area. In PAO BW, the permeate water is used for backwashing the membrane. Hence, it is necessary to make sure that the backwash cycle is as short as possible (Spiegler & Macleish 1981; Haidari et al. 2018).

Recently, Liberman (2017) (IDE Technologies Ltd, Israel) analyzed the three methods of osmotic backwash cleaning for RO membranes, that is direct (forward) osmosis cleaning (DOC), direct osmosis high salinity (DOHS), and pulse flow forward osmosis (PFFO) backwash. Among the three methods, the PFFO BW-enabled RO system is proven to protect the membrane from biofouling and scaling. The biofouling is mitigated by inducing a plasmolysis phenomenon in the microorganism. The scaling is controlled by sweeping the proto-crystals and minerals from the membrane surface during PFFO BW. The method proposed by IDE Technologies requires additionally a high pressure permeate pump and high concentrated saline water to enable PFFO BW in the RO system. To reduce the need of pump and saline water, Chatterjee et al. (2014) have designed a simplified PAO BW system by introducing an empty BW tank and solenoid valve after the RO membrane module to generate high pressure permeate for PAO BW.

In this study, to optimize the performance of the high recovery RO system for brackish water application, the impact of design and operating parameters on RO membrane performance is analyzed experimentally. Three different high recovery RO configurations have been explored to study their performance improvement by integrating the simplified PAO membrane cleaning process used in our previous patent application (Chatterjee et al. 2014). The design parameters considered in this study are channel dimension, spacer thickness, and number of leaves in a spiral-wound module. Operating parameters such as membrane recovery and PAO BW conditions are optimized to maximize the membrane life. The optimized design and operation conditions are used to design a high recovery RO system. The performance of the optimized high recovery RO system is evaluated by purifying pre-treated Brahmaputra river water. The RO-treated low total dissolved solids (TDS) water is used as feed water for demineralization (DM), such that DM plant softener media life is improved, treated water from the DM plant will be free from microplastics, and the hybrid high recovery RO + DM system should be able to reduce the water wastage and reduce overall water cost (Wenten et al. 2013). The use of nanofiltration membrane may be an alternative to RO membrane for low TDS feed water, but failed to reduce the TDS when reject is recycled in CCRO. Therefore, in this work, the RO membrane is used for water purification.

MATERIALS AND METHODS

High recovery system configurations

In this study, three different high recovery RO configurations are considered to identify the best suited system while integrating the proposed simplified PAO membrane BW concept. As shown in Figure 1, the performances of six different high recovery RO flowsheets are experimentally studied to quantify the cleaning efficiency of the PAO process. The list of equipment used in each of the flowsheet and operating conditions is described in Table 1. All six high recovery RO systems are designed to produce 15 LPH permeate flow rate at feed temperature 30 °C and feed pressure 5.2 bar. However, due to the variations in feed temperature and concentration, both permeate and feed flow rate is expected to vary, as given in Table 1. The flowsheet of system I is a simple low recovery RO process and most widely used in the small capacity RO membrane (75 GPD) based water purification system. However, to study and compare the effect of PAO BW cleaning at high recovery, the recovery of the system is fixed around 90%. A significant challenge in system I is the severe fouling caused by the reduced reject flow rate.

Table 1

High recovery RO system design and operating conditions

Flowsheet systems Feed
 
Membrane reject
 
Recycle flow (LPH) BW time (min) List of additional equipment used 
Q (LPH) V (m/s) Q (LPH) V (m/s) 
System I (D) 16.5 0.008 1.5 0.0008 NA NA Basic units 
System II (D) 40 0.020 25 0.0127 25 NA Basic units, recycle tank 
System III (D) 40 0.020 25 0.0127 23.5 NA Basic units, recycle tank, SV 
System III (S) 40 0.020 25 0.0127 23.5 NA Basic units, recycle tank, SV 
System IV (D) 16.5 0.008 1.5 0.0008 NA 6–8 Basic units, HPS, SV, timer, pressure tank 
System V (D) 40 0.020 25 0.0127 25 6–8 Basic units, HPS, SV, timer, pressure tank 
System V (S) 40 0.020 25 0.0127 25 6–8 Basic units, recycle tank, HPS, SV, timer, pressure tank 
System VI (D) 40 0.020 25 0.0127 23.5 6–8 Basic units, recycle tank, HPS, SV, timer, pressure tank 
Flowsheet systems Feed
 
Membrane reject
 
Recycle flow (LPH) BW time (min) List of additional equipment used 
Q (LPH) V (m/s) Q (LPH) V (m/s) 
System I (D) 16.5 0.008 1.5 0.0008 NA NA Basic units 
System II (D) 40 0.020 25 0.0127 25 NA Basic units, recycle tank 
System III (D) 40 0.020 25 0.0127 23.5 NA Basic units, recycle tank, SV 
System III (S) 40 0.020 25 0.0127 23.5 NA Basic units, recycle tank, SV 
System IV (D) 16.5 0.008 1.5 0.0008 NA 6–8 Basic units, HPS, SV, timer, pressure tank 
System V (D) 40 0.020 25 0.0127 25 6–8 Basic units, HPS, SV, timer, pressure tank 
System V (S) 40 0.020 25 0.0127 25 6–8 Basic units, recycle tank, HPS, SV, timer, pressure tank 
System VI (D) 40 0.020 25 0.0127 23.5 6–8 Basic units, recycle tank, HPS, SV, timer, pressure tank 

Q = flow and V = velocity.

Basic units: pump, pressure gauge, RO membrane, LPS, flow control valve (FCV).

Operating condition: feed pressure = 5.2 bar, BW pressure = 5.2 bar.

Designed permeate flowrate = ∼15 LPH. Note: S, single leaf, D, dual leaf.

Figure 1

Six different high recovery RO configurations. (a) System-I: Simple RO without BW (b) System-II: RO with Recycle Tank and without BW (Continuous purge) (c) System-III: RO with Recycle Tank and without BW (Discontinuous Purge) (d) System-IV: Simple RO with BW (e) System-V: RO with Recycle Tank and BW (Continuous Purge) (f) System-VI: RO with Recycle Tank and BW (Discontinuous Purge).

Figure 1

Six different high recovery RO configurations. (a) System-I: Simple RO without BW (b) System-II: RO with Recycle Tank and without BW (Continuous purge) (c) System-III: RO with Recycle Tank and without BW (Discontinuous Purge) (d) System-IV: Simple RO with BW (e) System-V: RO with Recycle Tank and BW (Continuous Purge) (f) System-VI: RO with Recycle Tank and BW (Discontinuous Purge).

Further, to overcome this issue, in system II, the membrane recovery is reduced, and overall system recovery is increased to ∼90% by recirculating the brine. The known quantity of brine was purged continuously from the recycle loop to prevent the membrane from severe fouling issues. In the flowsheet of system III, reject water is recirculated back to the recycle tank. When salt concentration in the recycle tank reaches a threshold level, the concentrated brine was purged from the system for a short period, to avoid salt deposition on the membrane surface. This flowsheet is also called a batch CCRO (Warsinger et al. 2016). In flowsheet systems I, II, and III, the PAO BW is introduced based on the author's previous patent application (Chatterjee et al. 2014). The corresponding BW introduced flowsheets are named as system IV, V, and VI, respectively.

The possibility of implementing IDE's PFO BW (Liberman 2017) in systems IV, V, and VI is also evaluated; as discussed in the introduction section, it can be implemented when system feed concentration is high enough (>15,000 ppm) to achieve target backwash flux. Therefore, the implementation of PFFO BW in the system with a low feed concentration (<1,000 ppm) is not viable, and PFO BW is implemented in this study. Consequently, the performance of both IDE's PFO BW and PFO BW process cannot be compared, and either one can be chosen for implementation with respect to feed concentration.

Effect of permeate channel dimensions vs membrane cleaning efficiency

The high-pressure drop in the permeate channel is expected to provide high BW flux next to the permeate collection tube and low BW flux at the dead-end region. Therefore, to overcome this issue, one has to design a permeate channel with minimized pressure drop. Further, to achieve it in a domestic RO membrane module, the identification of optimal permeate channel spacer and number of leaves is a crucial aspect.

Many researchers (Karode & Kumar 2001; Li et al. 2005; Song & Ma 2005; Ahmad & Lau 2006; Fimbresweihs & Wiley 2008; Wardeh & Morvan 2009; Kim & Elimelech 2012; Koutsou et al. 2013; Park & Kim 2013; She et al. 2013; Haidari et al. 2018) have studied the importance of the feed and permeate spacer dimensions to maximize the permeate flow rate keeping the pressure drop under control. The use of optimal spacer design parameters for both feed and permeate channel is a very important aspect for achieving optimal permeate flow. In general, 0.22–0.24 mm size permeate spacers are used for larger size modules, but 0.2–0.22 mm thickness permeate spacers are used in smaller modules (such as 75 and 100 GPD design capacity modules). Hence, there might be a significant pressure drop in the permeate channel, which affects the smaller module performance. Therefore, we used 0.22–0.28 mm thickness permeate spacer in this study. The product flow can be improved considerably by using a better spacer or a module with a shorter length and larger width. To study the effect of the membrane permeate channel dimension on the efficiency of PAO membrane cleaning process, the single leaf and dual leaf RO membrane modules have been used in system III and system V experiments. The specification of the membrane module used in this study is given in Table 2.

Table 2

Spiral-wound RO membrane specifications

Module parameters Module I Module II Module III 
Permeate channel length/permeate flow path distance (m) 1.372 0.686 1.372 
Permeate channel width/feed flow path distance (m) 0.224 0.224 0.224 
Feed spacer 
Thickness (mm) 0.40–0.43 0.40–0.43 0.40–0.43 
Material Poly ethylene Poly ethylene Poly ethylene 
Overlap angle (deg) 90 90 90 
Wales/inch 34–38 34–38 34–38 
Courses/inch 38–42 38–42 38–42 
Permeate spacer 
Thickness (mm) 0.22–0.28 0.22–0.28 0.40–0.43 
Material Bi component polyester Bi component polyester Poly ethylene 
Overlap angle (deg) 90 90 90 
Wales/inch 34–38 34–38 34–38 
Courses/inch 38–42 38–42 38–42 
Manufacturer DOW DOW DOW 
Tradename FILMTEC™ XLE FILMTEC™ XLE FILMTEC™ XLE 
No of leaves Single Dual Single 
Membrane surface area (m20.613 0.613 0.613 
Module parameters Module I Module II Module III 
Permeate channel length/permeate flow path distance (m) 1.372 0.686 1.372 
Permeate channel width/feed flow path distance (m) 0.224 0.224 0.224 
Feed spacer 
Thickness (mm) 0.40–0.43 0.40–0.43 0.40–0.43 
Material Poly ethylene Poly ethylene Poly ethylene 
Overlap angle (deg) 90 90 90 
Wales/inch 34–38 34–38 34–38 
Courses/inch 38–42 38–42 38–42 
Permeate spacer 
Thickness (mm) 0.22–0.28 0.22–0.28 0.40–0.43 
Material Bi component polyester Bi component polyester Poly ethylene 
Overlap angle (deg) 90 90 90 
Wales/inch 34–38 34–38 34–38 
Courses/inch 38–42 38–42 38–42 
Manufacturer DOW DOW DOW 
Tradename FILMTEC™ XLE FILMTEC™ XLE FILMTEC™ XLE 
No of leaves Single Dual Single 
Membrane surface area (m20.613 0.613 0.613 

Module I is a single leaf module with a thick spacer in the feed channel and thin spacer in the permeate channel. Module II is a dual leaf module with a thick spacer in the feed channel and thin spacer in the permeate channel. Module III is, again, a single leaf module having thick spacers in both feed and permeate channels. The module leaf length is adjusted to achieve an equal surface area (0.613 m2) for the entire modules used in this study.

For PAO BW, the optimized BW pressure and required water volume are fixed as per our previously reported experimental study (Chatterjee et al. 2014). The following steps are followed to fixed PAO BW parameters.

Step 1: During PAO BW, the pressure in the permeate channel is expected to be higher than the feed channel. The differential pressure at which the membrane separation layer could peel off was determined, and the operating PAO BW pressure was set at a pressure significantly less than that. For example, in this study, above 6.55 bar of PAO BW pressure, the membrane separation layer started peeling off from the support layer; therefore, the maximum BW pressure is fixed at 5.2 bar.

Step 2: In this step, the required BW water volume is calculated by conducting repeated experiments with respect to different water volume; we observed that ∼0.58 L is sufficient per PAO BW cycle.

Step 3: In this step, the required BW tank volume for a given backwash volume is calculated using ideal gas law: 
formula
(1)
where Vtank = volume of tank (l), PBW = BW pressure (bar), VWater = required water volume for BW, Patm = atmospheric pressure (1.013 bar). For example, to perform BW at 5.2 bar, the required tank volume will be 1.1959 times of water volume.

Step 4: PAO BW duration is fixed such that pressure in the BW tank declines to 1.013 bar from its initial pressure. The BW duration will vary with respect to feed concentration. In this study, it varied from 6 to 8 min.

Step 5: The frequency of the PAO BW can be optimized based on overall water recovery and membrane life. However, in this study, the PAO BW is triggered after every 5 L of permeate water production.

Normalization of membrane performance

In the RO process, both permeate flow and salt passage are significantly affected due to variations in feed water parameters such as temperature and feed concentration (Zhao & Taylor 2005). In this study, ASTM D4516 methods are used for the standardization of the permeate flow and salt passage data considering feed parameter variations. Thus, the measured actual RO permeate flow rate and concentration is standardized by using the ASTM D4516 method given in the Dow Technical Manual Excerpt (n.d.).

Membrane morphology

Membranes used in all six high recovery RO systems were analyzed using field emission scanning electron microscope (FESEM) (model: Gemini 300, make: Carl Zeiss) for membrane surface analysis and the samples were coated with gold to avoid charging. The virgin and fouled membrane surface compositions were analyzed using a FESEM-EDX instrument (make: Zeiss, model: Sigma), and the sample was double coated with gold to avoid charging. For FESEM observation, 3 kV accelerated voltage was used whereas 15 kV was used for EDX analysis. The membrane samples for EDX and FESEM analysis are taken from the middle of the membrane length and width. Also, membrane cross sections were obtained by breaking the membranes after freezing in liquid nitrogen. Membrane thickness is measured by using a digimatic outside micrometre (series: 293, make: Mitutoyo, Japan). The thickness of the fouling layer per liter of permeate water production is calculated according to Equation (2): 
formula
(2)
where tnewmembrane = thickness of the new membrane, tfouled membrane = thickness of the fouled membrane and Vpermeate = volume of permeate produced by fouled membrane.

High recovery RO experimental setup

Considering the large quantity of feed water requirements for testing of six different high recovery RO systems, pre-treated Brahmaputra river water (BRW) is used as feed water with SDI <3 and turbidity <2.1 NTU. The pre-treatment system includes coagulation, flocculation, clarification, settling, filtration, and disinfection of filter water. The pre-treated river water quality is given in Table 3. The RO-treated river water is used as feed water to the demineralization plant. The feed river water is passed through a sediment filter to remove more than 5 μm particles (Genpure sediment filter; Batch No: GP + 2WOR-5M and SL No: 055674). Feed water temperature and TDS were measured using a TDS meter (model No.: DM1, make: HM digital USA). Pre-treated water is pumped to a RO membrane booster pump from Hi-Tech Ltd, India (model No.: E-b 300) and TDS measurements from TDS meter are used in all six systems. The configuration details of RO membrane used in this study are given in Table 2 and modules are wound using a facility available at Eureka Forbes Ltd, India, as per the tabled specification.

Table 3

Pre-treated river water quality data

Parameter name Value 
Turbidity (NTU) 2.10 ± 0.10 
pH 6.60 ± 0.01 
Total iron (mg/L) 0.60 ± 0.01 
Chloride content (mg/L) 6.00 ± 0.01 
Total hardness (mg/L, CaCO356.67 ± 1.15 
Calcium (mg/L) 11.43 ± 0.05 
Sulfate content (mg/L) 18.14 ± 0.13 
Alkalinity (mg/L, CaCO352.67 ± 1.15 
Total coliform (organism/100 mL) MPN techniques 
Parameter name Value 
Turbidity (NTU) 2.10 ± 0.10 
pH 6.60 ± 0.01 
Total iron (mg/L) 0.60 ± 0.01 
Chloride content (mg/L) 6.00 ± 0.01 
Total hardness (mg/L, CaCO356.67 ± 1.15 
Calcium (mg/L) 11.43 ± 0.05 
Sulfate content (mg/L) 18.14 ± 0.13 
Alkalinity (mg/L, CaCO352.67 ± 1.15 
Total coliform (organism/100 mL) MPN techniques 

Experimental procedure

In this high recovery RO system, feed pressure (P1) is maintained at 5.2 bar. While in the PAO BW process, the BW pressure (P2) at the permeate side is maintained less than the actual operating feed pressure (P2 < P1). In the case of high BW pressure, the membrane separation layer may peel off from the support layer (Avraham et al. 2006; Sagiv et al. 2008).

System I is similar to the typical RO systems available in the Indian market for domestic drinking water applications. The performance of system I will be used to benchmark the performance of the other five configurations reported in this study. In the first stage, the coarser suspended sediments are removed from the feed water using pre-filter. Subsequently, a booster pump is used to maintain the feed pressure at 5.2 bar, which is connected to the RO membrane through a solenoid valve ‘S1’. The RO membrane removes the TDS and microplastics from feed water. RO reject and permeate is collected to measure volume and TDS. The RO feed flow/pressure is manipulated by using the pump speed controller.

System II is the modified version of system 1, where reject is recycled through the non-return valve (NRV), and that will increase the overall system recovery by decreasing membrane module recovery. This configuration is also called a fed-batch high recovery RO. To enable mixing between brine recycle and inlet feed water, a recycle tank is placed before the booster pump inlet. The continuous discharge of brine and permeate flow is collected in the storage tank to measure volume and TDS.

The process flow diagram of system III is similar to system II, but in system II the reject is purged continuously and in system III reject is purged discontinuously. Therefore, in system III, the reject line is connected through solenoid valve ‘S2’. The valve S2 will be opened when recycling tank TDS is greater than 800 ppm; then, concentrated reject water is purged through valve ‘S2’ until the TDS of recycle tank reaches feed concentration. This configuration is also called a batch high recovery RO.

The system IV process configuration is established by introducing a PAO BW process in system I. The BW tank is added after the RO membrane to collect the pressurized permeate water, and that will facilitate the PAO BW for membrane cleaning. Additionally, two solenoid valves (S2, S3), a high-pressure switch (HPS), and timer are used to enable automated PAO BW process. To introduce PAO BW, the air pressure in the BW tank is increased by closing the solenoid valve (S3) located after the BW tank, and this event is triggered during the RO normal operation. Since permeate water is pumped into the BW tank from the bottom end, the pressure in the tank will be increased by compressing the air present in the tank. When the tank pressure reaches the required PAO BW pressure (5.2 bar), then the pump will be switched off, the solenoid valve (S2) placed in the reject line will be opened, and the solenoid valve (S1) placed after the pump will be closed to enable PAO across permeate to feed channel. The BW tank pressure and PAO water flux will be decreasing from maximum BW pressure to minimum pressure. During the PAO BW process, the feed channel pressure is reduced to atmospheric pressure by opening the solenoid valve (S2).

System V is similar to the system II process flow diagram where PAO-assisted BW is supplied. Similarly, system VI is identical to system III, where PAO supported BW is supplied. The operating parameters such as pressure, flow, TDS of feed, permeate, reject and tank are continuously monitored every 10 minutes.

RESULTS AND DISCUSSION

Effect of SWRO module configuration parameter

The pressure drop in the permeate channel and its impact on product flow and efficiency of the PAO BW is analyzed by performing RO experiments using the membrane module reported in Table 2. The pure water permeability experiments are conducted at 30 °C feed conditions. As shown in Figure 2, module II and III are producing significantly higher flow rates compared to module I. Modules I, II, and III are producing 11.8 LPH, 13.4 LPH, and 14.1 LPH permeate flow, respectively, at a feed pressure of 5.2 bar.

Figure 2

Pure water permeability of different modules with respect to feed pressure.

Figure 2

Pure water permeability of different modules with respect to feed pressure.

The increased permeate flow is observed mainly due to reduced pressured drop in the permeate channel by decreasing the permeate channel length in module II and increased permeate flow area by introducing the thick permeate channel spacer in module II. However, at 5.2 bar feed pressure, the dual leaf module II can produce more permeate than the other two modules. This result concludes that modules II and III can provide high permeate flow and may be able to provide better PAO BW efficiency compared to module I. Similarly, a study on analysis of PAO BW efficiency is performed using module I and module II. The thick permeate spacer has increased the module III diameter more significantly than that of the other two modules. Further, the large diameter membrane module was forcing us to accommodate a bigger size pressure vessel in the domestic water purifier, and that was not viable. Therefore, module III is not used in PAO BW experiments.

As described in the previous section, the feed temperature is expected to vary due to seasonal variation in the river water temperature and feed TDS. The corresponding actual permeate flow and concentration concerning permeate volume for system V (D) is reported in Figure 3. The normalized permeate flow and concentration are calculated using ASTM 4156 method.

Figure 3

Normalized permeate flow and TDS for system V (D).

Figure 3

Normalized permeate flow and TDS for system V (D).

The following standard operating condition is used for normalization calculations: feed pressure 5.2 bar, feed concentration 100 mg/l, recovery 75% and feed temperature 25 °C. As shown in Figure 3, the deviation between actual and normalized data is observed as soon as actual operating conditions are deviating from the standard working state for system V (D). The normalized permeate flow for the system is found to be decreasing with the volume of product water produced. However, the rate of permeate flow decline is low for normalized data compared to actual data. This deviation is occurring mainly due to the temperature correction factor and variations in feed concentration. Similarly, the rate of increase in normalized permeate concentration is also found to be lower than that of actual data. Therefore, the normalization of the permeate flow and concentration is a crucial aspect while comparing the performance of two different membrane modules under different operating conditions. The actual permeate flow and actual TDS of all systems is shown in Figure S1 and S2 (in the Supplementary information).

The performances of modules I and II are shown in Figure 4. The dual leaf membrane module II is providing permeate water with low TDS, desired flow rate, and 84.23% total recovery up to 1,500 L of permeate water production. The single leaf membrane module has failed to work after 25 L of water production. The normalized permeate flow rate is decreasing with time for dual leaf module II, but for single leaf module I, the permeate flow rate is found to be increasing with time. However, the permeate flow rate of module I (∼12 LPH) is found to be lower than module II flow rate (∼17 LPH), and this is occurring due to a high pressure drop in the permeate channel. Also, the permeate quality deterioration leads to the conclusion that the membrane separation layer is damaged in the single leaf module. Therefore, in a short period of operation, the module I permeate TDS (normalized ∼20 ppm and actual ∼80 ppm) is found be higher than that of module II permeate TDS (normalized 15 ppm and actual ∼40 ppm). All the experiments are performed twice to confirm the membrane failure.

Figure 4

Performance of single and dual leaf modules with PAO backwash for sytem V. (a) Performance of single leaf membrane – module 1 (b) Performance of dual leaf membrane – module 2.

Figure 4

Performance of single and dual leaf modules with PAO backwash for sytem V. (a) Performance of single leaf membrane – module 1 (b) Performance of dual leaf membrane – module 2.

The performance of single and dual leaf membrane modules are evaluated without BW for system III. The permeate flow and TDS variation with the volume of product water produced are presented in Figure 5. Both modules are operated until the product flowrate reached 6 LPH. The rate of permeate flow rate decline is found to be similar for both modules. The permeate TDS is fluctuating due to discontinuous reject discharge for both modules.

Figure 5

Performance of single and dual leaf modules without PAO backwash for system III. (a) Performance of single leaf membrane – module 1 (b) Performance of dual leaf membrane – module 2.

Figure 5

Performance of single and dual leaf modules without PAO backwash for system III. (a) Performance of single leaf membrane – module 1 (b) Performance of dual leaf membrane – module 2.

Effect of PAO BW with respect to system configuration

Systems I and IV – simple RO system

As shown in Figure 6, the feed temperature is fluctuating due to sessional variations, that lead to fluctuation in the permeate flow rate and concentrations. Therefore, both permeate flow rate and concentration values are normalized as per ASTM D4516 method. For system I, the normalized permeate flow rate is continuously decreasing; but in system IV, the permeate flow initially decreased until 200 L of the permeate water is produced, and later the normalized permeate flow is stabilized at ∼14 LPH. However, due to PAO BW in system IV, the permeate TDS is fluctuating in between 5 and 40 ppm. The salt present in the permeate water accumulates on the permeate channel during PAO BW, and the accumulated salt is released when normal operation is initiated. The total recovery of system I (87%) is always higher than system IV (82.8%); this is mainly due to the loss of product water in the PAO BW process. However, system I has failed due to low permeate flow (9.1 LPH) after producing 790 L of permeate water and system IV can produce the permeate water with a flow rate of 13.80 LPH up to 1,500 L. The permeate flow rate is found to be stabilized at ∼14 LPH from 200 to 1,500 L; therefore, system IV is expected to continue its performance with time.

Figure 6

Effect of PAO backwash on simple RO configuration. (a) Performance of system I (b) Performance of system IV.

Figure 6

Effect of PAO backwash on simple RO configuration. (a) Performance of system I (b) Performance of system IV.

Systems II and V – RO with brine recycle and continues partial brine purge

In systems II and V, to achieve low fouling against the system I performance, brine is recycled to have high feed water cross-flow velocity across the membrane feed channel. Simultaneously, a partial amount of brine (∼1.5 ± 0.5 LPH) was purged continuously from the system to avoid the accumulation of foulants in the recycling loop. The continuous removal of concentrated brine will help to reduce the fouling rate on the membrane surface. As shown in Figure 7, the results conclude that for system II, the normalized permeate flow rate decreased from 20 LPH to 15 LPH within 100 L of permeate collection; later no variation is observed up to 1,500 L of permeate water production. However, in system V, up to 1,050 L, the normalized permeate flow rate was ∼15 LPH, then a sudden decline in permeate flow rate from 15 to 13 LPH is observed. The deviation in flow rate may occur due to sudden fluctuation in feed temperature (24 °C to 18 °C), feed and reject concentration (Figure S3 in the Supplementary information). In system II, permeate TDS is observed to be less than 1.5 ppm. Similar to system IV's permeate TDS observation, the system V permeate TDS is also fluctuating between 2 and 30 ppm. However, due to high feed cross-flow velocity, both systems II and V can produce permeate at the desired flow rate up to 1,500 L of permeate water collection. Permeate quality and flow rate is not decreased in system II, and hence, it is expected to continue its performance with time. Compared to previous systems, in system II and V, high recovery is observed, i.e., 88.21% and 84.23%, respectively. Systems II and V are found to perform better than systems I and IV for low feed TDS conditions. However, this has to be verified for higher feed TDS conditions.

Figure 7

Effect of PAO backwash on RO with reject cycle and continuous purge. (a) Performance of system II (b) Performance of system V.

Figure 7

Effect of PAO backwash on RO with reject cycle and continuous purge. (a) Performance of system II (b) Performance of system V.

Systems III and VI – RO with brine recycle and discontinuous brine purge

Systems III and VI's performance are presented in Figure 8. The performance of system III is found to be better than system I; however, compared to system II, system III is not performing satisfactorily. The system III product flow rate is decreased from 17.5 to 7.5 LPH within 500 L of permeate collection. The poor performance observed may be due to the accumulation of foulants with complete brine recirculation and discontinuous brine discharge. However, the performance of system VI is found to be superior to all other reported systems regarding consistent permeate flow rate and TDS up to 1,500 L of permeate collection without compromising overall system water recovery (89.53%). In system VI, the concentration of feed is increasing with complete brine recycle (∼80 ppm to 700 ppm), and it will be increasing until the PAO BW is enabled for every 5 L of pure water production. The increased concentration difference between feed and permeate increases the net PAO driving force. This phenomenon will improve the PAO BW flux across the feed channel, and that leads to providing better membrane cleaning. Further, it can be justified from measured backwash water TDS (Figure S4 in the Supplementary information section), i.e., the backwash water TDS for systems V and VI is found to be higher than system IV).

Figure 8

Effect of PAO backwash on RO with reject cycle and discontinuous discharge. (a) Performance of system III (b) Performance of system VI.

Figure 8

Effect of PAO backwash on RO with reject cycle and discontinuous discharge. (a) Performance of system III (b) Performance of system VI.

The system VI permeate flow rate was observed as ∼14 LPH. However, as described earlier, due to PAO BW, the system VI permeate TDS is fluctuating between 15 and 50 ppm. Particularly after BW, TDS is found to be high and then decreases with time. This occurring phenomenon may be due to the accumulation of solute in the permeate channel during PAO BW and then it is released during normal operation. The reason for the accumulation of salt in the permeate channel during the BW is due to the rejection of the membrane skin during this stage.

The rejection takes place in both directions, also in the BW stage, and the salts accumulate in the support layer. It is released back immediately after return to regular operation.

The summary of all six systems' performance is given in Table 4, and Figures S1 and S2 in the Supplementary information. Systems I, III, and V are not able to provide a designed flow rate (10 to 15 LPH with respect to feed water quality variation), so those systems are not suitable for the high recovery operation. Among the others, system IV has used less specific energy consumption (NSEC). However, its fouling resisting ability has to be tested for higher feed water TDS conditions. The next best systems concerning SEC are systems VI and II. Similar results are observed from membrane thickness measurement due to fouling. As described in the previous section, due to complete brine recirculation, the high fouling thickness is observed in system III. On the other hand, low fouling is observed in system VI, mainly due to improved PAO-assisted BW flux that was enabled by an additional high concentration difference across the permeate and feed channel. System V is performing better than system II up to 1,000 L of water production, and later due to reducing feed temperature in system V, system II is performing better than system V (Figure 7); the reduced temperature may also lead to inadequate BW flux and cleaning. The effect of temperature on BW flux will be explored in a future study. Systems III and VI are operated under fully recycle (batch) mode, and that leads to the existence of high concentration contaminants in the recycling loop; therefore, in fully recycle mode systems, severe fouling is expected compared to other systems.

Table 4

Summary of system performance for all six systems

 Pure water permeate flow rate
 
Permeate properties
 
Total NSEC tfouling 
New After use Flow rate at the start Flow rate at the end TDS at end Total 
System L/h L/h L/h L/h ppm flow L recovery % kWh/m3 nm 
Ia 20.06 11.59 19.60 9.10 21 790 87.05 0.194 8.99 
II 18.63 13.22 19.50 13.60 1,500 88.21 0.747 6.47 
IIIa 17.50 10.10 16.77 7.40 13 510 96.30 1.866 34.31 
IV 20.19 14.30 20.45 13.85 32 1,500 82.88 0.183 5.40 
Va 18.00 12.43 16.34 9.04 30 1,503 84.23 0.533 4.66 
VI 18.20 14.25 14.55 11.16 25 1,450 89.53 0.577 3.79 
 Pure water permeate flow rate
 
Permeate properties
 
Total NSEC tfouling 
New After use Flow rate at the start Flow rate at the end TDS at end Total 
System L/h L/h L/h L/h ppm flow L recovery % kWh/m3 nm 
Ia 20.06 11.59 19.60 9.10 21 790 87.05 0.194 8.99 
II 18.63 13.22 19.50 13.60 1,500 88.21 0.747 6.47 
IIIa 17.50 10.10 16.77 7.40 13 510 96.30 1.866 34.31 
IV 20.19 14.30 20.45 13.85 32 1,500 82.88 0.183 5.40 
Va 18.00 12.43 16.34 9.04 30 1,503 84.23 0.533 4.66 
VI 18.20 14.25 14.55 11.16 25 1,450 89.53 0.577 3.79 

New membrane thickness = 140.5 μm.

Total recovery % = Total permeate flow *100/Total feed flow.

aSystem final condition is concluded as failed either due to insufficient product flow rate or high permeate TDS.

Effect of PAO BW on membrane surface

At the end of each experiment, the membrane was analyzed with FESEM-EDX analysis to determine the surface structure changes and composition of both virgin membrane and used membranes (systems I to VI). FESEM top view and side view images are shown in Figure 9. While comparing new membrane with all six systems' used membrane, the image shows that salt deposition on membrane surface is 7.1, 9.7, 17.5, 8.1, 7.0, and 5.5 μm from system 1 to VI, respectively. This result replicates the observation made based on normalized membrane performance data.

Figure 9

SEM image of new and used RO membranes to study changes in membrane thickness.

Figure 9

SEM image of new and used RO membranes to study changes in membrane thickness.

Further to capture the fouling deposition along the membrane permeate channel length, the used spiral-wound RO modules are unfolded, and photo images are shown in Figure 10. In systems IV, V, and VI, small colour variations were observed along the length of the leaf, even though the photograph could not adequately reflect those observations. Further to confirm the same, the EDX analysis of membrane samples along the length of the permeate channel is performed. The EDX results also conclude that more element deposition is observed near the permeate channel dead-end compared to the other end, and this is mainly due to low BW flux at the dead-end of the permeate channel. This may be due to less foulant deposition in the membrane near the permeate collection tube, compared to the permeate dead-end. The fouling layer in systems IV and V is observed with amorphous agglomerates on top of the membrane. The system VI membrane surface image is found to be similar to the new membrane image. The measured elemental composition of the membrane surface taken from the center of the module is presented in Table 5.

Table 5

EDX analysis results for membrane surface composition

Membranes based on the system no. Elemental composition (Wt %)
 
Ca Si Zn Fe Al Others 
Unused 76.10 13.80 5.30 2.50 – – – – – 1.70 
System I (D) 46.70 32.43 5.80 5.00 4.73 1.00 0.93 0.53 0.60 2.27 
System II (D) 26.53 42.96 3.10 1.37 1.63 9.30 3.13 4.30 4.83 2.85 
System III (D) 32.36 43.90 2.53 0.00 13.66 1.70 4.43 0.10 0.20 1.12 
System III (S) 25.00 61.90 0.20 1.80 0.15 4.80 – 0.15 4.35 1.65 
System IV (D) 32.40 56.15 0.40 3.80 0.10 3.00 – 0.25 2.80 1.10 
System V (S) 54.16 33.43 2.96 1.60 0.40 3.50 0.46 1.13 1.90 0.46 
System V (D) 68.30 19.50 2.90 8.10 – 0.45 0.25 0.15 0.10 0.25 
System VI (D) 79.20 13.80 5.06 1.13 – – 0.10 – – 0.71 
Membranes based on the system no. Elemental composition (Wt %)
 
Ca Si Zn Fe Al Others 
Unused 76.10 13.80 5.30 2.50 – – – – – 1.70 
System I (D) 46.70 32.43 5.80 5.00 4.73 1.00 0.93 0.53 0.60 2.27 
System II (D) 26.53 42.96 3.10 1.37 1.63 9.30 3.13 4.30 4.83 2.85 
System III (D) 32.36 43.90 2.53 0.00 13.66 1.70 4.43 0.10 0.20 1.12 
System III (S) 25.00 61.90 0.20 1.80 0.15 4.80 – 0.15 4.35 1.65 
System IV (D) 32.40 56.15 0.40 3.80 0.10 3.00 – 0.25 2.80 1.10 
System V (S) 54.16 33.43 2.96 1.60 0.40 3.50 0.46 1.13 1.90 0.46 
System V (D) 68.30 19.50 2.90 8.10 – 0.45 0.25 0.15 0.10 0.25 
System VI (D) 79.20 13.80 5.06 1.13 – – 0.10 – – 0.71 
Figure 10

Images of the unfolded membrane module.

Figure 10

Images of the unfolded membrane module.

In the virgin membrane, the elements belonging to the thin film composite membrane such as C, O, S, and N are detected. The inorganic elements present in raw feed water such as Ca, Si, Zn, Fe, and Al are not detected in the new membrane. However, Ca, Si, Zn, and Fe are detected in the membrane that is used in systems 1–VI. Compared to all other membranes, the membrane used in system VI is detected to have less inorganic colloid deposition. Once again, this observation concludes that PAO BW is more effective in system VI (D) compared to other systems. From the overall system analysis, the lesser inorganic colloid deposition is observed for a system operated with PAO BW. In systems II and III, even though brine recirculation enables reduced concentration polarization and improved water flux, in another aspect, the reject recycle may result in accumulation of foulants (contaminants) in the membrane feed. This resulted in more deposition containing an element of Ca, Si, Fe, and Al in systems II and III compared with system I.

CONCLUSION

The performance of three different configurations of high recovery RO systems is analyzed regarding the PAO membrane BW process. The experimental results conclude that spiral-wound module with less pressure drop in the permeate channel can provide efficient PAO membrane BW. Further, the use of concentrated feed solution during PAO BW will improve the BW efficiency, and it is achieved by using a concentrated brine. Among the three different high recovery RO systems, the system with brine recirculation (CCRO) is observed to have less membrane fouling than the others. However, CCRO uses more equipment than the others, which leads to higher capital cost. From another aspect, a system without brine recycle (single stage RO) is found to be more energy efficient than the other systems reported in this work, and this has to be verified for higher feed concentration. The effectiveness of PAO BW in RO membrane is counterbalanced, to some extent, by the significantly higher permeate TDS when backwash is applied.

ACKNOWLEDGEMENTS

The Department of Chemical Engineering and Central Instruments Facility, Indian Institute of Technology Guwahati, Assam is also acknowledged for providing access to FESEM, EDX, and digimatic outside micrometer instruments.

REFERENCES

REFERENCES
Astudillo
C.
,
Parra
J.
,
González
S.
&
Cancino
B.
2010
A new parameter for membrane cleaning evaluation
.
Sep. Purif. Technol.
73
,
286
293
.
doi:10.1016/j.seppur.2010.04.015
.
Avraham
N.
,
Dosoretz
C.
&
Semiat
R.
2006
Osmotic backwash process in RO membranes
.
Desalination
199
,
387
389
.
doi:10.1016/j.desal.2006.03.088
.
Chatterjee
J.
,
Pratap
S.
&
Subbiah
S.
2014
A Water Purifier and A Process of Cleaning the Membrane
.
Pub. No.: WO/2016/066382, International Application No.: PCT/EP2015/073029, Publication date: 06.05.2016, International filing date: 06.10.2015
.
Dow
n. d.
Technical Manual Excerpt, S. Report, FILMTEC Membranes System Operation: Record Keeping, doi:Form No. 609-02084-0905
.
Fimbresweihs
G.
&
Wiley
D.
2008
Numerical study of two-dimensional multi-layer spacer designs for minimum drag and maximum mass transfer
.
J. Memb. Sci.
325
,
809
822
.
doi:10.1016/j.memsci.2008.09.005
.
Gal
Z.
&
Efraty
A.
2016
CCD series no. 18: record low energy in closed-circuit desalination of ocean seawater with nanoH2O elements without ERD
.
Desalin. Water Treat.
57
,
9180
9189
.
doi:10.1080/19443994.2015.1035500
.
Gao
L. X.
,
Rahardianto
A.
,
Gu
H.
,
Christofides
P. D.
&
Cohen
Y.
2016
Novel design and operational control of integrated ultrafiltration – reverse osmosis system with RO concentrate backwash
.
Desalination
382
,
43
52
.
doi:10.1016/j.desal.2015.12.022
.
Haidari
A. H.
,
Heijman
S. G. J.
&
van der Meer
W. G. J.
2018
Optimal design of spacers in reverse osmosis
.
Sep. Purif. Technol.
192
,
441
456
.
doi:10.1016/j.seppur.2017.10.042
.
Kim
Y. C.
&
Elimelech
M.
2012
Adverse impact of feed channel spacers on the performance of pressure retarded osmosis
.
Environ. Sci. Technol.
46
,
4673
4681
.
doi:10.1021/es3002597
.
Kim
J. O.
,
Jung
J. T.
,
Yeom
I. T.
&
Aoh
G. H.
2007
Effect of fouling reduction by ozone backwashing in a microfiltration system with advanced new membrane material
.
Desalination
202
,
361
368
.
doi:10.1016/j.desal.2005.12.075
.
Koutsou
C. P.
,
Karabelas
A. J.
&
Goudoulas
T. B.
2013
Characteristics of permeate-side spacers of spiral wound membrane modules
.
Desalination
322
,
131
136
.
doi:10.1016/j.desal.2013.05.015
.
Li
F.
,
Meindersma
W.
,
De Haan
A. B.
&
Reith
T.
2005
Novel spacers for mass transfer enhancement in membrane separations
.
J. Memb. Sci.
253
,
1
12
.
doi:10.1016/j.memsci.2004.12.019
.
Liberman
B.
2009
Direct osmosis cleaning. US7563375 B2
.
Liberman
B.
2017
Three methods of forward osmosis cleaning for RO membranes
.
Desalination
431
,
22
26
.
doi:10.1016/j.desal.2017.11.023
.
Lu
X.
,
Kujundzic
E.
,
Mizrahi
G.
,
Wang
J.
,
Cobry
K.
,
Peterson
M.
,
Gilron
J.
&
Greenberg
A. R.
2012
Ultrasonic sensor control of flow reversal in RO desalination-Part 1: mitigation of calcium sulfate scaling
.
J. Memb. Sci.
419–420
,
20
32
.
doi:10.1016/j.memsci.2012.05.027
.
Luján-Facundo
M. J.
,
Mendoza-Roca
J. a.
,
Cuartas-Uribe
B.
&
Álvarez-Blanco
S.
2013
Ultrasonic cleaning of ultrafiltration membranes fouled with BSA solution
.
Sep. Purif. Technol.
120
,
275
281
.
doi:10.1016/j.seppur.2013.10.018
.
Mizrahi
G.
,
Wong
K.
,
Lu
X.
,
Kujundzic
E.
,
Greenberg
A. R.
&
Gilron
J.
2012
Ultrasonic sensor control of flow reversal in RO desalination. Part 2: mitigation of calcium carbonate scaling
.
J. Memb. Sci.
419–420
,
9
19
.
doi:10.1016/j.memsci.2012.05.026
.
Motsa
M. M.
,
Mamba
B. B.
,
Thwala
J. M.
&
Verliefde
A. R. D.
2017
Osmotic backwash of fouled FO membranes: cleaning mechanisms and membrane surface properties after cleaning
.
Desalination
402
,
62
71
.
doi:10.1016/j.desal.2016.09.018
.
Porcelli
N.
&
Judd
S.
2010
Chemical cleaning of potable water membranes: a review
.
Sep. Purif. Technol.
71
,
137
143
.
doi:10.1016/j.seppur.2009.12.007
.
Qin
J.-J.
,
Liberman
B.
&
Kekre
K. A.
2009
Direct osmosis for reverse osmosis fouling control: principles, applications and recent developments
.
Open Chem. Eng. J.
3
,
8
16
.
doi:10.2174/1874123100903010008
.
Qin
J. J.
,
Oo
M. H.
,
Kekre
K. A.
&
Liberman
B.
2010
Development of novel backwash cleaning technique for reverse osmosis in reclamation of secondary effluent
.
J. Memb. Sci.
346
,
8
14
.
doi:10.1016/j.memsci.2009.08.011
.
Qiu
T.
&
Davies
P. A.
2012
Comparison of configurations for high-recovery inland desalination systems
.
Water (Switzerland)
4
,
690
706
.
doi:10.3390/w4030690
.
Ramon
G.
,
Agnon
Y.
&
Dosoretz
C.
2010
Dynamics of an osmotic backwash cycle
.
J. Memb. Sci.
364
,
157
166
.
doi:10.1016/j.memsci.2010.08.008
.
Ramon
G. Z.
,
Nguyen
T. V.
&
Hoek
E. M. V.
2013
Osmosis-assisted cleaning of organic-fouled seawater RO membranes
.
Chem. Eng. J.
218
,
173
182
.
doi:10.1016/j.cej.2012.12.006
.
Sagiv
A.
&
Semiat
R.
2005
Backwash of RO spiral wound membranes
.
Desalination
179
,
1
9
.
doi:doi.org/10.1016/j.desal.2004.11.050
.
Sagiv
A.
&
Semiat
R.
2010a
Parameters affecting backwash variables of RO membranes
.
Desalination
261
,
347
353
.
doi:10.1016/j.desal.2010.04.012
.
Sagiv
A.
&
Semiat
R.
2010b
Modeling of backwash cleaning methods for RO membranes
.
Desalination
261
,
338
346
.
doi:10.1016/j.desal.2010.04.054
.
Sagiv
A.
,
Avraham
N.
,
Dosoretz
C. G.
&
Semiat
R.
2008
Osmotic backwash mechanism of reverse osmosis membranes
.
J. Memb. Sci.
322
,
225
233
.
doi:10.1016/j.memsci.2008.05.055
.
Salvador Cob
S.
,
Beaupin
C.
,
Hofs
B.
,
Nederlof
M. M.
,
Harmsen
D. J. H.
,
Cornelissen
E. R.
,
Zwijnenburg
A.
,
Genceli Güner
F. E.
&
Witkamp
G. J.
2012
Silica and silicate precipitation as limiting factors in high-recovery reverse osmosis operations
.
J. Memb. Sci.
423–424
,
1
10
.
doi:10.1016/j.memsci.2012.07.016
.
Septon
J.
&
Efraty
A.
2016
CCD series no-17: application of the BWRO-CCD technology for high-recovery low-energy desalination of domestic effluents
.
Desalin. Water Treat.
57
,
9585
9601
.
doi:10.1080/19443994.2015.1035498
.
She
Q.
,
Hou
D.
,
Liu
J.
,
Tan
K. H.
&
Tang
C. Y.
2013
Effect of feed spacer induced membrane deformation on the performance of pressure retarded osmosis (PRO): implications for PRO process operation
.
J. Memb. Sci.
445
,
170
182
.
doi:10.1016/j.memsci.2013.05.061
.
Spiegler
K. S.
&
Macleish
J. H.
1981
Molecular (osmotic and electro-osmotic) backwash of cellulose acetate hyperfiltration membranes
.
J. Memb. Sci.
8
,
173
192
.
doi:10.1016/S0376-7388(00)82089-X
.
Stover
R. L.
2013
Industrial and brackish water treatment with closed circuit reverse osmosis
.
Desalin. Water Treat.
51
,
1124
1130
.
doi:10.1080/19443994.2012.699341
.
Trägårdh
G.
1989
Membrane cleaning
.
Desalination
71
,
325
335
.
doi:10.1016/0011-9164(89)85033-7
.
Uchymiak
M.
,
Bartman
A. R.
,
Daltrophe
N.
,
Weissman
M.
,
Gilron
J.
,
Christofides
P. D.
,
Kaiser
W. J.
&
Cohen
Y.
2009
Brackish water reverse osmosis (BWRO) operation in feed flow reversal mode using an ex situ scale observation detector (EXSOD)
.
J. Memb. Sci.
341
,
60
66
.
doi:10.1016/j.memsci.2009.05.039
.
Warsinger
D. M.
,
Tow
E. W.
,
Nayar
K. G.
,
Maswadeh
L. A.
&
Lienhard
V. J. H.
2016
Energy efficiency of batch and semi-batch (CCRO) reverse osmosis desalination
.
Water Res.
106
,
272
282
.
doi:10.1016/j.watres.2016.09.029
.
Wenten
I. G.
,
Khoiruddin
K.
,
Arfianto
F.
&
Zudiharto
,
2013
Bench scale electrodeionization for high pressure boiler feed water
.
Desalination
314
,
109
114
.
doi:10.1016/j.desal.2013.01.008
.
Zhao
Y.
&
Taylor
J. S.
2005
Assessment of ASTM D 4516 for evaluation of reverse osmosis membrane performance
.
Desalination
180
,
231
244
.
doi:10.1016/j.desal.2004.11.089
.

Supplementary data