Reverse osmosis brine treatment using direct contact membrane distillation: Effects of feed temperature and velocity

Desalination xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Desalination journal homepage: www.elsevier.com/locate/desal

Reverse osmosis brine treatment using direct contact membrane distillation: Effects of feed temperature and velocity Zhongsen Yan, Haiyang Yang, Fangshu Qu⁎, Huarong Yu, Heng Liang, Guibai Li, Jun Ma State Key Laboratory of Urban Water Resource and Environment (SKLUWRE), Harbin Institute of Technology, Harbin 150090, PR China

A R T I C L E I N F O

A B S T R A C T

Keywords: Membrane distillation (MD) Reverse osmosis (RO) brine Membrane scaling Membrane wetting

Membrane distillation (MD) is a promising technology for reverse osmosis (RO) brine treatment due to its superb tolerance to high salinity. In this work, a hydrophobic PVDF membrane was applied to treat simulated RO brine under different feed temperatures and velocities with membrane flux, permeate conductivity and thermal performance monitored. The raw membrane was characterized with respect to contact angle, porosity, thickness, maximum pore sizes, liquid entry pressure and clean water flux. The fouled membrane was characterized by scanning electron microscope (SEM) coupled with an energy dispersive X-ray spectroscopy (EDX). The results showed that MD achieved excellent desalination performance with the permeate conductivity lower than 11 μS cm− 1 and recovery rate higher than 70% during the RO brine treatment. Almost no membrane scaling was observed in the initial stage, but significant scaling occurred due to overconcentration (concentration factor > 3.3) of RO brine, resulting in serious pore wetting and decreased thermal performance. Moreover, increasing feed velocity was helpful to alleviate the membrane scaling, but the desalination performance would be impaired to some extents with feed velocity exceeding 0.4 m s− 1. The increasing feed temperature could significantly increase the membrane flux, but membrane scaling was accelerated resulting in a lower recovery.

1. Introduction With growing application of reverse osmosis (RO) for seawater desalination, there exists a high quantity of RO brine with high salinity [1]. Discharging the RO brine directly into offshore water may result in serious coastal environment pollution and land salinization due to the high salinity of RO brine and several chemical agents added to pretreat seawater prior to RO [2–4]. In the meantime, discharging RO brine may also cause a waste of water resources leading to a higher burden of associated facilities [5]. Therefore, to treat and to recycle RO brine is of great importance for seawater desalination industry. A variety of technologies were investigated to treat RO brine, including pressure-driven membrane processes, current-driven membrane processes and thermal-driven processes [6,7]. The pressuredriven membrane processes, such as RO and nanofiltration, were able to produce high quality water. However, a much bigger energy demand resulting from concentration polarization and membrane fouling constricts the application of the aforementioned pressure driven processes. The current-driven membrane processes like electro-osmosis have the same advantages, but the energy consumption is proportional to the brine salinity and the rate of salt removal [8]. Hence, both pressuredriven and current-driven membrane processes are not economically

⁎

feasible in treating high-salinity RO brine [6]. In contrast, thermaldriven processes possess a great potential in the brine treatment due to theirs insensitivity to the high salinity [9]. Solar evaporation is commonly used for the brine disposal at a low cost [10,11], but the large land demand for evaporation pond restricts its application, especially in some locations with slow evaporation rates [10]. Besides, leakage of evaporation ponds may pose a high contamination potential to groundwater [12]. Membrane distillation (MD) is a thermal-driven separation process combined with membrane technology. In the MD process, volatile compounds evaporate on the feed side and are collected on the other side of membrane. According to the steam collection form, MD can mainly be classified into direct contact membrane distillation (DCMD), air gap membrane distillation (AGMD), sweep gas membrane distillation (SGMD) and vacuum membrane distillation (VMD) [13]. DCMD, which has the simplest configuration, is widely used in the scientific investigation of MD [14]. Compared to traditional evaporation processes, MD is able to be operated within a relatively moderate temperature range (40 °C to 80 °C), which makes it far more compatible with low-grade waste heat sources or renewable energy [13]. In addition, membrane modules can be arranged compactly due to small vapor space demand, considerably reducing the footprint of desalination

Corresponding author. E-mail address: [email protected] (F. Qu).

http://dx.doi.org/10.1016/j.desal.2017.09.010 Received 15 May 2017; Received in revised form 5 August 2017; Accepted 11 September 2017 0011-9164/ © 2017 Published by Elsevier B.V.

Please cite this article as: Yan, Z., Desalination (2017), http://dx.doi.org/10.1016/j.desal.2017.09.010

Desalination xxx (xxxx) xxx–xxx

Z. Yan et al.

Cooling

plants [15]. Because the vapor pressure of feed is not as strongly dependent as osmotic pressure, a relatively high recovery rate can be expected in MD process. Therefore, MD becomes attractive for RO brine treatment in recent years. Qu et al. [16] concentrated the primary RO brine by 40 times using DCMD, and achieved a water recovery rate as high as 98.8%. Mericq et al. [5] have performed a study on the seawater reverse osmosis brine treatment using VMD, and an overall recovery factor of 89% was achieved by coupling RO and VMD via a simulation analysis. Despite great desalination performance presented in literature, MD has not been extensively applied for industrial desalination [17]. It is considered that low energy efficiency and lack of commercially available and high performance membranes constrict the commercialization of MD [18]. However, as the development of multi-stage MD, energy efficiency or gained output ratio (GOR) in MD was comparable to other thermal-driven technologies (multistage flash, MSF; multiple effect evaporation) [19]. In addition, low-grade thermal energy and renewable energy can be used in MD, which will effectively decrease the operation cost of energy (close to or lower than RO) [19,20]. With regard to commercial MD membranes, plenty of excellent and highperformance membranes are fabricated in the lab recently [21–23]. Therefore, MD is promising for the industrial application in the near future. Similar to other membrane process, membrane scaling is another factor substantially constricting the application of MD. Given the growing number of studies on the scaling in other membrane processes (such as NF, RO), scaling of sparingly soluble salt is yet to be fully understood in a high-concentration and high-temperature environment adjacent to hydrophobic MD membranes [17,24]. The deposition of pollutants and crystals on the membrane surface may lead to severe wetting, resulting in permeate deterioration [25,26]. Moreover, scaling will also affect heat conduction and thermal performance, but it is hardly analyzed. Therefore, it is significant to investigate the scaling and fouling phenomena in the MD process. In this study, DCMD was performed to treat a simulated RO brine containing a small account of sparingly soluble salt (calcium sulfate). Pure water was used to investigate the effects of feed velocity and temperature on clean water flux of the MD membrane. Heat transport analysis was conducted to study the temperature polarization and thermal performance in the MD module. RO brine treatment using DCMD was performed to investigate the efficiency of MD at a series of feed temperatures and velocities. Membrane fouling and wetting were monitored during the whole process. Membrane scaling was characterized with respect to morphology and reversibility.

Pump T

T

Feed tank

T

Membrane module

Permeate tank

T Electronic scales Heating Pressure sensor

Level controller

T Concentrate tank

Thermocouple

Pump

Fig. 1. DCMD experimental setup.

self-manufactured experimental setup of which the schematic diagram is shown in Fig. 1. The DCMD system was comprised of a feed circulation system, a permeate circulation system and a membrane module. In the feed circulation system, 200 mL feed was added to the concentration tank beforehand, and raw water was supplemented by a feed tank to maintain a constant water volume in the feed circulation. Feed was continuously pumped to go through a heat exchanger and the membrane module before returning to the concentration tank using a peristatic pump (WT3000-1JB, Longer, China). The heat exchanger was adopted to maintain the water temperature of feed at 50 ± 1 °C. In the permeate circulation system, 200 mL of Milli-Q deionized water was initially added to serve as a cold medium. The permeate was circulated through a cooling unit and the membrane module using a peristatic pump. A thermostat bath (DC-0510, Scientz, Ningbo, China) was applied to maintain the permeate temperature at 20 ± 1 °C. A flat sheet MD membrane with an effective area 25 cm2 was placed inside the membrane cell with a plastic mesh used as the support. The feed and permeate in both sides of the MD membrane were circulated in the opposite direction. The permeate velocity was fixed at 0.25 m s− 1, and the feed velocity varied from 0.05 to 0.4 m s− 1. Inside the cell, the feed evaporated in the feed side, and then the vapor penetrated through membrane pores due to the vapor pressure gradient induced by the temperature difference. Finally, the vapor condensed on the permeate side. The permeate tank was placed on an electronic balance connected to a computer with weighing data automatically logged every 5 min. To predict pore wetting, a conductivity meter (Seven Compact S230, Mettler Toledo, Switzerland), which was connected to the computer, was installed to monitor the permeate conductivity on line with an electrode (InLab 741, Mettler Toledo, Switzerland). In addition, the hydraulic pressure of feed at the cell inlet was also monitored using a pressure transducer (PTP708, Tuopo Electric, Foshan, China). Because the circulation flow rate was maintained constant, an increase in feed pressure could manifest the membrane scaling which would reduce the cross-section area for the circulated water flow.

2. Materials and methods 2.1. Membranes and feed water A hydrophobic PVDF membrane (IPVH00010), which was purchased from Millipore Corp (Millipore, MA, USA), was used in this study. The PVDF membranes were characterized with respect to contact angle, porosity, maximum pore size and thickness beforehand. The chemical agents, including sodium chloride (NaCl), potassium chloride (KCl), magnesium sulfate (MgSO4), magnesium chloride hexahydrate (MgCl2·6H2O) and calcium chloride (CaCl2), were purchased from Sinopharm Chemical Reagent Corp (Beijing, China). The simulated RO brine was prepared in lab containing 32.61 g/L NaCl, 1.03 g/L KCl, 6.96 g/L MgCl2·6H2O, 4.50 g/L MgSO4 and 4.50 g/L CaCl2 according to the composition of RO brine given by Ge et al. [27]. The mixed solution was filtered by a 0.45 μm microfiltration membrane (Taoyuan, China) subsequently. The conductivity of the feed was determined as 62 ± 1 mS cm− 1.

2.3. Membrane characterization Membrane was pre-dried in the vacuum drying oven (70 °C and − 0.08 MPa) for one day before characterization. Contact angle was measured by an optical contact angle meter (JYSP-180, Jinshengxin Co., Ltd., China) to evaluate the hydrophobicity of the MD membrane. A square piece of the MD membrane (0.5 cm × 0.5 cm) was fixed on a slide glass which was placed on the sample stage of the device. 3 μL water was dripped on the membrane surface at room temperature (25 °C) using a microsyringe. The images of water drop were captured in 3 s and then the contact angle could be obtained using a three-point method. The measurement was repeated for 6 times. Liquid entry pressure of water (LEPw) was used to evaluate the tolerance of MD membrane to pore wetting. A piece of disk membrane (diameter 25 mm) was fixed on a self-made filtration cell (as shown in Fig. S1) to determine the LEPw. Detailed instruction of measurement

2.2. DCMD experiment The RO brine treatment by DCMD was conducted in a bench-scale 2

Desalination xxx (xxxx) xxx–xxx

Z. Yan et al.

2dh, s ⎞0.5 Nu = kdc 0.664Re 0.5 Pr 0.33 ⎛ ⎝ lm ⎠ ⎜

⎟

(6)

where 0.086 df −0.039 0.75 ⎛ θ kdc = 1.654 ⎛ ⎞ φ sin ⎛ ⎞ ⎞ ⎝ ⎝ 2 ⎠⎠ ⎝H⎠ ⎜

⎟

⎜

⎟

(7)

Here, kdc is the correction factor for spacer geometry, dh,s is the hydraulic diameter for the spacer-filled channel, df is spacer filament size, lm is the mesh size, H is the spacer thickness, φ is the spacer voidage, and θ is the hydrodynamic angle. Heat transfer coefficient of membrane can be calculated from the thermal conductivities of the hydrophobic membrane polymer (km) and air trapped inside the membrane pores (kg) [29].

hm =

(1 − ε ) km + k g ε

At steady state, the heat transfers in three regions have same values according to the first law of thermodynamics.

Fig. 2. Schematic diagram of heat and mass transfer through the membrane in DCMD configuration [14].

Qp = Qf = Qm

TPC =

(2)

(11)

Qf = Cf mf (Tf , in − Tf , out )

(12)

3. Results and discussions 3.1. Characteristics of the MD membrane Table 1 shows various characteristics of the raw PVDF membrane used in this study. According to the manufacturer, the nominal pore size of the membrane was 0.45 μm. The maximum pore size (0.91 μm) was two times bigger than the nominal pore size. The contact angle of the membrane was determined as 132°, indicating strong hydrophobicity of the membrane. The CA value for the same membrane varied from 124.2° to 140° as reported in literature [32,33]. Moreover, it can be observed that the MD membrane exhibited a LEPw of 150 kPa, which

(3)

where hf, hp and hm are the heat transfer coefficient of feed, permeate and membrane, respectively; Tf and Tp are the average inlet and outlet water temperatures at feed and permeate sides, respectively; Tm,f and Tm,p are the water temperature at the feed and permeate sides of membrane surface, respectively; J is the permeate flux, and HV is the enthalpy of the vapor which can be calculated by Eq. (4) [30].

Hv {T } = 1.7535(T ) + 2024.3

(4)

Table 1 Characteristics of the MD membrane used in experiments ( ± indicates the standard errors, n ≥ 3).

−1

where H is in kJ kg , and T in K. Heat transfer coefficient of feed and permeate can be calculated through the definition of Nusselt (Nu).

h=

Nu k δ

(10)

where Qf and Qc represent heat transfer in the feed channel of the membrane module and heat conduction of membrane respectively, Cf is specific heat, mf is mass flux of feed (calculation shown in Supplementary information, S4), and Tf,in and Tf,out are the inlet and outlet temperature of MD module in the feed side respectively.

Heat transfer through the membrane (Qm) can be described by Eq. (3):

Tm, f + Tm, p ⎫ + hm (Tm, f − Tm, p) Qm = JHv ⎧ ⎬ ⎨ 2 ⎭ ⎩

Tf − Tp

JHv JHv = Qin JHv + Qc

GOR = The heat transfer involved in DCMD can be divided into three regions as shown in Fig. 2. Since the effect of mass transfer on the heat transfer in the feed and permeate boundary layers was negligible [29], heat transfer at the feed side Qf and permeate side Qp can be calculated by Eqs. (1) and (2).

Qp = hp (Tm, p − Tp)

Tm, f − Tm, p

In order to evaluate the thermal performance of MD module, GOR (gained output ratio, without counting the pumping energy) was calculated by Eqs. (11) and (12) [31].

2.4. Heat transport analysis

(1)

(9)

Combining Eqs. (1)–(3) and (9), membrane/feed interface temperature Tm,f and membrane/permeate interface temperature Tm,p can be figured out. So the temperature polarization coefficient (TPC) can be calculated by Eq. (10).

was presented in the Supplementary information. Maximum pore size of membrane was determined using the bubble point method [28]. Membrane surface morphology and membrane scaling were characterized by an environmental scanning electron microscopy (ESEM, FEI Quanta 200FEG, USA). The membrane sample was pre-dried and coated with gold before ESEM analysis. Energy Dispersive X-ray Spectroscopy (EDX) was used to identify the composition of the scaling composition on the membrane surface.

Qf = hf (Tf − Tm, f )

(8)

δ

(5)

Net-type spacer was used in the module to support the membrane and enhance the turbulence. So Nu calculation was referenced to Eq. (7) [30]. 3

Parameters

Value

Pore size (μm) Average thickness (μm) Maximum pore size (μm) Contact angle (°) LEPw (kPa) Porosity (%)

0.45 113 ± 3 0.91 ± 0.02 132.04 ± 3.06 150 ± 4 72.73 ± 1.51

Desalination xxx (xxxx) xxx–xxx

Z. Yan et al.

(a) 100

(a)

100% Flux

TPC

1.0

80%

80

Normalized flux

Flux (L m-2h-1)

0.8 60% TPC

60

40%

40

0.6 40 °C 50 °C 60 °C

0.4

Initial membrane flux 11.56 L m-2 h-1 20.57 L m-2 h-1 41.25 L m-2 h-1

20%

20

0.2 0%

0 30

40

50 Temperature (°C)

60

(b) 50

0.0

70

1

2

3 CF

4

5

4

5

100% Flux

(b) 1.0

TPC

40

80%

20

40%

10

20%

TPC

60%

Normalized flux

Flux (L m-2h-1)

0.8

30

0.6 Initial membrane flux 0.1 m s-1 14.90 L m-2 h-1 0.25 m s-1 20.57 L m-2 h-1 0.4 m s-1 22.99 L m-2 h-1

0.4

0.2 0 0.0

0.1

0.2 0.3 0.4 Velocity of feed side (m s-1)

0% 0.5

0.0 1

Fig. 3. Characteristic of MD membrane in different feed water temperature (feed water velocity 0.25 m s− 1) (a) and feed water velocity (feed water temperature 50 °C) (b) using pure water.

2

3 CF

Fig. 4. Variations of normalized flux in different feed temperatures (feed velocity: 0.25 m s− 1) (a) and feed velocities (feed temperature: 50 °C) (b) in RO brine treatment. Membrane flux was normalized by the initial membrane flux.

was much higher than the feed pressure applied, revealing a low risk of pressure-induced pore wetting. To get more insights into the characteristics of the MD membrane, the clean water flux was determined at a series of feed temperatures and velocity. As shown in Fig. 3, the membrane flux was only 4.4 L m− 2 h− 1 at the lowest feed temperature (30 °C). As the feed temperature grew to 70 °C, the membrane flux increased to 72.3 L m− 2 h− 1. This is attributed to the great increase in vapor pressure at feed side which drove the mass transport through the MD membrane. It can be also observed that the membrane flux increased with the feed velocity, especially in the feed velocity range of 0.05 to 0.1 m s− 1. The probable reason is that the increased feed velocity intensifies the turbulence, resulting in less temperature polarization at the feed side of DCMD. As shown in Fig. 3(b), TPC increased with feed velocity, especially within the range from 0.05 to 0.1 m− 1, verifying the constriction of temperature polarization due to the higher feed velocity applied. In another study on a AGMD system, a slight increase in flux had ever been reported with increasing feed velocity [34,35]. Compared with AGMD, the DCMD in this study was more apparently affected by the temperature polarization since the cold water was directly used as the cold medium. Alklaibi et al. [36] compared two types of MD processes by modeling, and found that the flux of DCMD was much more sensitive to the increase in feed velocity than that of AGMD.

tendency of flux variation with concentration factor (CF) was observed irrespective of operational conditions. When the CF was smaller than 3.3, the flux all decreased at a slow rate. It was considered that membrane scaling and vapor gradient decrease due to concentration were two significant culprits for the flux reduction [37]. According to the components given in Section 2.1, feed contained a large amount of calcium and sulfate. Since the solubility of calcium sulfate was affected by the ion strength and temperature, an aqueous geochemical calculations software (PHREEQC, version 3.3) was used to calculate the saturation index (SI) of calcium sulfate. There were two types of crystallizations of calcium sulfate (Anhydrite and Gypsum), as shown in the Fig. S2. When the CF increased exceeding 2.6, calcium sulfate was likely to crystallize due to the oversaturation at all the three feed temperatures applied. However, the decreasing rate of membrane flux was stable and insignificant and no substantial increase of feed pressure was observed in this stage (shown in Fig. S3), revealing no significant scaling occurred. To verify this assumption, morphological observation was performed with the membranes sampled at different CF levels (feed velocity: 0.25 m s− 1, feed temperature: 50 °C, CF = 1.5, 2, 2.5 and 3). As shown in SEM images (Fig. S4), no visible scaling occurred on the membrane surface. In addition, the amount of calcium was rare and not regularly varied with CF (as shown in Table S2). It was considered that the occurrence of calcium sulfate crystallization needed sufficiently induction time [38], so membrane scaling was excluded from the main reason for the membrane flux reduction in the primary stage. The variation of vapor pressure with the increasing CF was calculated via Raoult's law (shown in Table S1 in the Supplementary information)

3.2. Effects of feed temperature and velocity on membrane flux variation Membrane flux during RO brine treatment by DCMD was investigated within a feed temperature range from 40 to 60 °C and a velocity range from 0.1 to 0.4 m s− 1. As shown in Fig. 4, the similar 4

Desalination xxx (xxxx) xxx–xxx

Z. Yan et al.

(a)

[39]. The decrease in vapor pressure was < 4% as the CF increased to 3.3, while the membrane flux was reduced by 15%. This result indicates the decrease of vapor pressure induced by the concentration was obviously smaller than membrane flux reduction, even if concentration polarization was accounted during calculation. Therefore, there are some other factors involved in membrane flux variation. It was reported that the viscosity of NaCl solution increased by 30%, when the concentration increased from 5 wt% to 15 wt% at 45 °C [40]. In this work, the viscosity of the RO brine increased by 14% when the concentration was triple (shown in Fig. S5). The majority of the flux decline in this stage may be attributed to the increasing viscosity of concentrate which would decrease the turbulence and increase temperature polarization [40,41]. It can also observed in Fig. 4 that sharp decreases in DCMD flux occurred irrespective of feed temperature and velocity when the CF exceeded 3.3. Compared with Fig. S3, a rapid increase in feed pressure occurred at the same time of sharp decrease of flux, revealing that the reduction in DCMD flux was closely associated with salt crystallization which resulted in growth of scaling layer on the membrane surface. Specifically, the sharp flux decrease was firstly observed in the scenario where the highest feed temperature (60 °C) was applied. By contrast, the flux decline was alleviated to some extent at the lower feed temperature with the final specific flux of 0.41 at the feed temperature of 40 °C. It can be observed in Fig. 4(b) that the flux decrease was slightly alleviated with increasing feed velocity. Both feed temperature and velocity only played a minor role in scaling. It was considered that supersaturation and sufficient induction time were the two significant factors for calcium sulfate crystallization. The supersaturation of calcium sulfate was reached when the CF exceeded 2.6 in this study. The increasing temperature might reduce the induction time, resulting in accelerated membrane scaling and thus flux decrease when a higher feed temperature was applied [26,38]. Moreover, the concentration polarization, which might also contribute to membrane scaling and flux decline, could be alleviated as higher feed velocity was applied to impose stronger turbulence adjacent to membrane surface. Therefore, the sharp flux decrease was delayed to some degree when the feed velocity increased from 0.1 m s− 1 to 0.4 m s− 1. Overall, the DCMD flux would substantially decrease when membrane scaling occurred and the application of lower feed temperature and higher feed velocity might slightly alleviate flux decline.

50 40 °C

50 °C

60 °C

Conductivity (µs/cm)

40

30

20

10

0 1

2

3 CF

4

5

(b) 50 0.1 m s-1

0.25 m s-1

0.4 m s-1

Conductivity (µs/cm)

40

30

20

10

0 1

2

3 CF

4

5

Fig. 5. Effect of feed temperature (feed velocity: 0.25 m s− 1) (a) and feed velocity (feed temperature: 50 °C) (b) on permeate conductivity variation in RO brine treatment.

3.3. Effects of feed temperature and velocity on permeate conductivity

except when the lowest feed temperature was applied. Several studies claimed that the crystallization on the membrane surface could cause severe wetting of membrane pores and significantly increase permeate conductivity [40,43]. Although the feed pressure increased quickly due to the scaling of membrane (shown in Fig. S3), it was still far lower than the LEPw value of the new MD membrane. Therefore, membrane scaling was suspected to change the membrane properties which significantly reduced the resistance of wetting. As membrane wetting was closely affected by the scaling, permeate conductivity was similarly affected by the feed temperature and velocity as DCMD flux. It can be observed in Fig. 5(a) that the increasing feed temperature accelerated the occurrence of pore wetting with the sharp increase of permeate conductivity occurring at the CF of 4.5, 3.6, and 3.6 when the feed temperatures were 40, 50 and 60 °C, respectively. This is consistent with the result reported by Nguyen et al. that pore wetting could be delayed at low feed temperature [42]. In addition, MD pore wetting was postponed by the highest velocity in small degree. Overall, the feed velocity should be properly design to constrict membrane partial wetting and concentration factor need be careful concerned according to the salinity of RO brine to avoid significant wetting.

Fig. 5 shows the conductivity profiles during MD operated under different feed temperatures and velocities. It can be observed that the desalination performance of MD was superb with the permeate conductivity lower than 11 μS cm− 1 during RO brine treatment. However, the total rejection of salts was hard to achieve and the rejection was impacted by feed temperature and velocity to some degree. In the initial stage (CF < 3.3), the feed temperature showed minor impact to permeate conductivity, but the desalination performance would be impaired by increasing feed velocity from 0.1 m s− 1 to 0.4 m s− 1 with permeate conductivity exceeding 10 μS cm− 1 at the CF of 3. It was considered that there were a few defects on membrane surface which enable mild partial wetting to occur [27]. As is known, the hydraulic pressure at the inlet of a MD system usually increases with the velocity. In this study, the feed pressures at the cell inlet were about 10, 3 and 1 kPa under the velocity of 0.40, 0.25, 0.10 m s− 1, respectively (shown in Fig. S3). Compared with the LEPw of the MD membrane (150 kPa), feed pressure was unable to wet the membrane. Hence, the minor increase in permeate conductivity was attributed to the partial wetting which was exacerbated at the higher feed velocity. When the CF increased exceeding 3.3, the sharp increase in permeate conductivity could be observed, indicating the occurrence of pore wetting [42]. Compared with Fig. 4, it can be note that pore wetting occurred almost simultaneously with sharply decrease in flux

3.4. Effects of feed temperature and velocity on GOR Fig. 6 shows the GOR variations with CF during RO brine treatment 5

Desalination xxx (xxxx) xxx–xxx

Z. Yan et al.

demonstrated that the GOR of the single-stage MD was theoretically < 1 without heat recovery [31]. As shown in Fig. 6, the GOR of DCMD increased with the increasing feed temperature and velocity. This is mainly due to more rapid increase of membrane flux with feed temperature than that of heat lost. As shown in the Section 3.2, the membrane flux exponentially increased with the feed temperature, whereas the heat lost only linearly increased with the temperature gradient between two sides of the MD membrane. Therefore, the ineffective heat lost via conduction played a less significant role at higher feed temperatures, resulting in the increased GOR [19]. With regard to feed velocity, the temperature adjacent to the membrane surface was increased by reducing temperature polarization, leading to increased GOR at higher feed velocities during RO brine treatment using DCMD. It can be also observed that the GOR decreased with CF in a similar pattern with the flux. In the initial stage, GOR slowly decreased with increasing CF. Because the vapor pressure slightly decreases with feed salinity, more heat is demanded for water vaporization in the MD process at higher CF levels, resulting in decreased GOR values with CF [31]. Similar to membrane flux, GOR also sharply decreased irrespective of feed temperature and velocity, when significant membrane scaling occurred as CF exceeded 3.3. The result implies that the thermal performance of DCMD was substantially impaired in the presence of scaling layer during RO brine treatment. This is attributed to the sharply-decreased flux in the presence of scaling layer as shown in Fig. 4. Overall, the thermal efficiency of DCMD increased with both feed temperature and velocity during RO brine treatment, and the membrane scaling should be controlled to maintain the thermal performance by restricting CF below 3.3.

(a) 1.0 40 °C

50 °C

60 °C

0.8

GOR

0.6

0.4

0.2

0.0 1

2

3 CF

4

5

(b) 1.0 0.1 m s-1

0.25 m s-1

0.4 m s-1

0.8

GOR

0.6

0.4

3.5. Characterization of membrane surface and scaling layer 0.2

To verify the membrane scaling during RO brine treatment, the fouled membranes were characterized using SEM after filtration tests. As the SEM images of membrane surface for different scenarios were similar due to the identical feed water used, only one group of SEM image and EDX spectrum, obtained at the feed temperature and velocity of 50 °C and 0.25 m s− 1, respectively, were presented and compared together with those of the new membrane. It can be observed in Fig. 7 that there were plenty of pores on the raw membrane, whereas the fouled membrane was covered by a thick scaling layer, indicating the occurrence of membrane scaling. The scaling layer could not only block the transport of vapor, but also increase the heat resistance between the bulk solution and membrane surface, resulting in decreased vapor gradient and MD flux [40]. As shown in the EDX spectra of the scaling

0.0 1

2

3 CF

4

5

Fig. 6. Effects of feed temperature (feed velocity: 0.25 m s− 1) (a) and feed velocity (feed temperature: 50 °C) (b) on GOR in RO brine treatment using DCMD.

using DCMD. It can be observed that the GOR in the MD process was generally < 0.7, significantly lower than the GOR values of existing thermal technologies such as MED or MSF (varying from 6 to 10) [44]. This was due to a single-stage MD system used in this work. It was

Fig. 7. SEM images of the virgin membrane (a) and fouled membrane (b).

6

Desalination xxx (xxxx) xxx–xxx

Z. Yan et al.

Fig. 8. EDX spectra of fouled membrane (feed temperature and velocity were 50 °C and 0.25 m s− 1 respectively).

4. Conclusion

layer (Fig. 8), there were strong peaks of calcium, sulfate and oxygen, revealing that sparingly soluble calcium sulfate was the main component of the deposit on the membranes surfaces. Ge et al. [27] had also reported that CaSO4 crystals formed and covered the membrane surface during RO brine treatment using DCMD. Moreover, the calcium sulfate scaling was also reported in the real RO brine treatment [45]. To remove the scaling layer, the MD membrane was flushed with 200 mL pure water for 20 min. The flux of the flushed membrane was determined with the raw water to evaluate the flux reversibility. As show in the Fig. 9, the flux reversibility was varying from 88% to 96%, revealing that the scaling layer could be effectively removed by flushing. However, the flux reversibility of the MD membranes decreased with the increasing feed temperature. The probable reason is that the scaling layer on the MD membrane was less compact and more removable under lower feed temperature [26]. Moreover, the solubility of calcium sulfate decreased with increasing water temperature, resulting in relatively compact scaling layer at the higher feed temperature.

In this work, RO brine treatment using DCMD was investigated, and the impacts of feed temperature and velocity on membrane permeability, membrane scaling pore wetting and thermal performance were systematically studied. The following conclusions can be drawn. 1. MD was able to produce the pure water (conductivity: lower than 11 μS cm− 1) from the high concentrate RO brine (conductivity: 61 mS cm− 1) with a recovery rate higher than 70%, but significant scaling occurred with the concentration factor exceeding 3.3. 2. The MD flux significantly increased with increasing feed temperature due to greater vapor gradient, but the membrane scaling was aggravated at the higher feed temperature, resulting in accelerated sharp decrease in MD flux and reduced water recover rate. 3. Increasing feed velocity will increase membrane flux to some extent by reducing temperature polarization, but the risk of partial pore wetting was exacerbated at higher velocity with the occurrence of sharp increase in permeate conductivity accelerated. 4. The thermal efficiency of DCMD increased with both feed temperature and velocity during RO brine treatment but was substantially reduced in the occurrence of membrane scaling. 5. The scaling layer on the MD membrane could be easily removed by flushing with the flux recovered by > 88%. The reversibility of MD flux decreased with increasing feed temperature, probably attributed to more compact scaling layer formed under higher feed temperature.

Flux reversibility

100%

80%

Acknowledgements 60%

This work was financially supported by Open Project of State Key Laboratory of Urban Water Resource and Environment (ES201511-02) and Fund from Heilongjiang Postdoctoral Fund (Grants LBH-TZ1612). Special thanks should be given to Prof. Mark R. Wiesner who gave innovative guidance on the application of MD for wastewater treatment.

40%

20% 40

50

60

Appendix A. Supplementary data

Temperature (°C) Fig. 9. Effect of feed temperature on flux reversibility in RO brine treatment (feed temperature and velocity were 50 °C and 0.25 m s− 1 respectively).

Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.desal.2017.09.010. 7

Desalination xxx (xxxx) xxx–xxx

Z. Yan et al.

References [24]

[1] V.G. Gude, Desalination and sustainability – an appraisal and current perspective, Water Res. 89 (2016) 87–106. [2] S. Lattemann, T. Höpner, Environmental impact and impact assessment of seawater desalination, Desalination 220 (2008) 1–15. [3] D.G. Randall, J. Nathoo, A succinct review of the treatment of reverse osmosis brines using freeze crystallization, J. Water Process Eng. 8 (2015) 186–194. [4] D. Qu, J. Wang, L. Wang, D. Hou, Z. Luan, B. Wang, Integration of accelerated precipitation softening with membrane distillation for high-recovery desalination of primary reverse osmosis concentrate, Sep. Purif. Technol. 67 (2009) 21–25. [5] J. Mericq, S. Laborie, C. Cabassud, Vacuum membrane distillation of seawater reverse osmosis brines, Water Res. 44 (2010) 5260–5273. [6] C.R. Martinetti, A.E. Childress, T.Y. Cath, High recovery of concentrated RO brines using forward osmosis and membrane distillation, J. Membr. Sci. 331 (2009) 31–39. [7] J.M. Arnal, M. Sancho, I. Iborra, J.M. Gozálvez, A. Santafé, J. Lora, Concentration of brines from RO desalination plants by natural evaporation, Desalination 182 (2005) 435–439. [8] Y. Zhang, K. Ghyselbrecht, B. Meesschaert, L. Pinoy, B.V.D. Bruggen, Electrodialysis on RO concentrate to improve water recovery in wastewater reclamation, Fuel Energy Abstr. 378 (2011) 101–110. [9] L.F. Greenlee, F. Testa, D.F. Lawler, B.D. Freeman, P. Moulin, Effect of antiscalant degradation on salt precipitation and solid/liquid separation of RO concentrate, J. Membr. Sci. 366 (2011) 48–61. [10] A. Pérez-González, A.M. Urtiaga, R. Ibáñez, I. Ortiz, State of the art and review on the treatment technologies of water reverse osmosis concentrates, Water Res. 46 (2012) 267–283. [11] O. Lefebvre, R. Moletta, Treatment of organic pollution in industrial saline wastewater: a literature review, Water Res. 40 (2006) 3671–3682. [12] L. Katzir, Y. Volkmann, N. Daltrophe, E. Korngold, R. Mesalem, Y. Oren, J. Gilron, WAIV - wind aided intensified evaporation for brine volume reduction and generating mineral byproducts, Desalin. Water Treat. 13 (2010) 63–73. [13] B.B. Ashoor, S. Mansour, A. Giwa, V. Dufour, S.W. Hasan, Principles and applications of direct contact membrane distillation (DCMD): a comprehensive review, Desalination 398 (2016) 222–246. [14] M. Khayet, Membranes and theoretical modeling of membrane distillation: a review, Adv. Colloid Interf. Sci. 164 (2011) 56–88. [15] L. Mariah, C.A. Buckley, C.J. Brouckaert, E. Curcio, E. Drioli, D. Jaganyi, D. Ramjugernath, Membrane distillation of concentrated brines—role of water activities in the evaluation of driving force, J. Membr. Sci. 280 (2006) 937–947. [16] D. Qu, J. Wang, B. Fan, Z. Luan, D. Hou, Study on concentrating primary reverse osmosis retentate by direct contact membrane distillation, Desalination 247 (2009) 540–550. [17] L.D. Tijing, Y.C. Woo, J. Choi, S. Lee, S. Kim, H.K. Shon, Fouling and its control in membrane distillation—a review, J. Membr. Sci. 475 (2015) 215–244. [18] P. Wang, T. Chung, Recent advances in membrane distillation processes: membrane development, configuration design and application exploring, J. Membr. Sci. 474 (2015) 39–56. [19] M. Khayet, Solar desalination by membrane distillation: dispersion in energy consumption analysis and water production costs (a review), Desalination 308 (2013) 89–101. [20] U.K. Kesieme, N. Milne, H. Aral, C.Y. Cheng, M. Duke, Economic analysis of desalination technologies in the context of carbon pricing, and opportunities for membrane distillation, Desalination 323 (2013) 66–74. [21] Y. Huang, Q. Huang, H. Liu, C. Zhang, Y. You, N. Li, C. Xiao, Preparation, characterization, and applications of electrospun ultrafine fibrous PTFE porous membranes, J. Membr. Sci. 523 (2017) 317–326. [22] J. Zuo, S. Bonyadi, T. Chung, Exploring the potential of commercial polyethylene membranes for desalination by membrane distillation, J. Membr. Sci. 497 (2016) 239–247. [23] J. Wang, L. Zheng, Z. Wu, Y. Zhang, X. Zhang, Fabrication of hydrophobic flat sheet

[25]

[26] [27] [28] [29] [30]

[31]

[32]

[33]

[34]

[35]

[36] [37]

[38] [39]

[40]

[41]

[42] [43]

[44] [45]

8

and hollow fiber membranes from PVDF and PVDF-CTFE for membrane distillation, J. Membr. Sci. 497 (2016) 183–193. D.M. Warsinger, J. Swaminathan, E. Guillen-Burrieza, H.A. Arafat, J.H. Lienhard V, Scaling and fouling in membrane distillation for desalination applications: a review, Desalination 356 (2015) 294–313. E. Guillen-Burrieza, A. Servi, B.S. Lalia, H.A. Arafat, Membrane structure and surface morphology impact on the wetting of MD membranes, J. Membr. Sci. 483 (2015) 94–103. L.D. Nghiem, T. Cath, A scaling mitigation approach during direct contact membrane distillation, Sep. Purif. Technol. 80 (2011) 315–322. J. Ge, Y. Peng, Z. Li, P. Chen, S. Wang, Membrane fouling and wetting in a DCMD process for RO brine concentration, Desalination 344 (2014) 97–107. E. Jakobs, W.J. Koros, Ceramic membrane characterization via the bubble point technique, J. Membr. Sci. 124 (1997) 149–159. M. Qtaishat, T. Matsuura, B. Kruczek, M. Khayet, Heat and mass transfer analysis in direct contact membrane distillation, Desalination 219 (2008) 272–292. J. Phattaranawik, R. Jiraratananon, A.G. Fane, Heat transport and membrane distillation coefficients in direct contact membrane distillation, J. Membr. Sci. 212 (2003) 177–193. H.W. Chung, J. Swaminathan, D.M. Warsinger, J.H. Lienhard V, Multistage vacuum membrane distillation (MSVMD) systems for high salinity applications, J. Membr. Sci. 497 (2016) 128–141. X. Wang, Z. Wang, Z. Wang, Y. Cao, J. Meng, Tethering of hyperbranched polyols using PEI as a building block to synthesize antifouling PVDF membranes, Appl. Surf. Sci. 419 (2017) 546–556. X. Lu, Y. Peng, H. Qiu, X. Liu, L. Ge, Anti-fouling membranes by manipulating surface wettability and their anti-fouling mechanism, Desalination 413 (2017) 127–135. H.C. Duong, A.R. Chivas, B. Nelemans, M. Duke, S. Gray, T.Y. Cath, L.D. Nghiem, Treatment of RO brine from CSG produced water by spiral-wound air gap membrane distillation — a pilot study, Desalination 366 (2015) 121–129. J. Xu, Y.B. Singh, G.L. Amy, N. Ghaffour, Effect of operating parameters and membrane characteristics on air gap membrane distillation performance for the treatment of highly saline water, J. Membr. Sci. 512 (2016) 73–82. A.M. Alklaibi, N. Lior, Comparative study of direct-contact and air-gap membrane distillation processes, Ind. Eng. Chem. Res. 46 (2007) 584–590. Y. Yun, R. Ma, W. Zhang, A.G. Fane, J. Li, Direct contact membrane distillation mechanism for high concentration NaCl solutions, Desalination 188 (2006) 251–262. M. Prisciandaro, A. Lancia, D. Musmarra, Gypsum nucleation into sodium chloride solutions, AICHE J. 47 (2001) 929–934. G. Dong, J.F. Kim, J.H. Kim, E. Drioli, Y.M. Lee, Open-source predictive simulators for scale-up of direct contact membrane distillation modules for seawater desalination, Desalination 402 (2017) 72–87. P. Zhang, P. Knötig, S. Gray, M. Duke, Scale reduction and cleaning techniques during direct contact membrane distillation of seawater reverse osmosis brine, Desalination 374 (2015) 20–30. S. Meng, Y. Ye, J. Mansouri, V. Chen, Fouling and crystallisation behaviour of superhydrophobic nano-composite PVDF membranes in direct contact membrane distillation, J. Membr. Sci. 463 (2014) 102–112. Q.S. Lee, S. Lee, Fouling analysis and control in a DCMD process for SWRO brine, Desalination 367 (2015) 21–27. F. He, K.K. Sirkar, J. Gilron, Studies on scaling of membranes in desalination by DCMD: CaCO3 and mixed CaCO3/CaSO4 systems, Chem. Eng. Sci. 64 (2009) 1844–1859. J. De Gunzbourg, D. Larger, Cogeneration applied to very high efficiency thermal seawater desalination plants, Desalination 125 (1999) 203–208. H.C. Duong, M. Duke, S. Gray, P. Cooper, L.D. Nghiem, Membrane scaling and prevention techniques during seawater desalination by air gap membrane distillation, Desalination 397 (2016) 92–100.