Influence of operating conditions on the removal of metals and sulfate from copper acid mine drainage by nanofiltration

Influence of operating conditions on the removal of metals and sulfate from copper acid mine drainage by nanofiltration

Chemical Engineering Journal 345 (2018) 114–125 Contents lists available at ScienceDirect Chemical Engineering Journal journal homepage: www.elsevie...

2MB Sizes 0 Downloads 19 Views

Chemical Engineering Journal 345 (2018) 114–125

Contents lists available at ScienceDirect

Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej

Influence of operating conditions on the removal of metals and sulfate from copper acid mine drainage by nanofiltration L. Pinoa, C. Vargasa, A. Schwarzb, R. Borqueza, a b

T



Chemical Engineering Department, Universidad de Concepción, P.O Box 160-C, Concepción, Chile Civil Engineering Department, Universidad de Concepción, P.O Box 160-C, Concepción, Chile

H I GH L IG H T S

G R A P HI CA L A B S T A R CT

spiral wound membranes de• Both monstrated a high metal and sulfate removal capacity.

difference in rejections was ob• No served at high pressures in both membranes.

Reynolds number does not • Increase significantly affect the permeate flow. measurements provide important • VRF information for scaling up AMD treatment.

shows excellent performance • NF270 until ten days of continuous operation.

A R T I C LE I N FO

A B S T R A C T

Keywords: Acid mine drainage Nanofiltration Spiral-wound membranes Heavy metals Resistance-in-series Membrane fouling

The primary objective of this investigation was to evaluate the ability of two commercial spiral-wound membranes (NF90 and NF270) to remove metals and sulfate from acid mine drainage from an active copper mine. The structural and surface properties of the membranes, hydrodynamic conditions, polarization, and filtration resistance had a significant influence on the effectiveness of the treatment. The obtained results demonstrated a good removal capacity in both membranes (> 94%) at a low operating pressure (15 bar). The increase in pressure had a strong impact on permeate flux, concentration polarization, and contaminant removal rate; however, the increase in the value of the Reynolds number exhibited had no significant effects. The NF270 membrane was selected to perform concentration and long-term tests as it demonstrated a high treatment capacity, high rejections, low resistance, and low polarization at moderate pressures. The permeate flux in the concentration and continuous operation tests decreased by ∼45% and ∼12%, respectively, due to the increases in resistance of ∼63% and ∼13%, respectively, while rejection exhibited a slight increase in both tests. The model parameters successfully identified the convective and diffusive contributions to ions transport in all experiments.

1. Introduction The problem of sulfide mineral oxidation associated with acid mine drainage (AMD) has been an alarming environmental issue for the



Corresponding author. E-mail address: [email protected] (R. Borquez).

https://doi.org/10.1016/j.cej.2018.03.070 Received 9 January 2018; Received in revised form 11 March 2018; Accepted 13 March 2018 Available online 15 March 2018 1385-8947/ © 2018 Elsevier B.V. All rights reserved.

mining industry worldwide [1]. Although the chemical composition of AMD depends on the climatological and geological conditions of the mining zone, AMD is commonly characterized by low pH and high concentrations of sulfate, heavy metals (Fe, Cu, Pb, Zn, Cd, Co), and

Chemical Engineering Journal 345 (2018) 114–125

L. Pino et al.

Nomenclature Symbol Δhyd G∘

Am CRi Cbi Cmi Cpi Di Jw R 0i RF RF + P

RI RM RP

RT Ri ki kw ri Δr n AMD P T V VRF v

Description (Units) standard molar Gibbs energy of solvation or hydration of ions (kJ mol−1) effective filtration area (m2) concentration of solute i in the rejected component (mg L−1) concentration of solute i in the feed (mg L−1) concentration of solute i on the membrane surface (mg L−1) concentration of solute i in the permeate (mg L−1) diffusion coefficient (cm2 s−1) permeate water flux (L h−1 m−2) observed rejection (%) mass transfer resistance due to fouling (m−1) mass transfer resistance due to fouling and polarization (m−1) mass transfer resistance due to scaling (m−1) mass transfer resistance due to the membrane (m−1) mass transfer resistance due to polarization (m−1)

total resistance (m−1) real (intrinsic) rejection of solute i (%) mass transfer coefficient (cm2/s) hydraulic permeability constant (L h−1 m−2 bar−1) ion radius (nm) hydrated ion layer width (nm) number of water molecules in the hydrated layer (–) acid mine drainage (–) pressure (bar) temperature (°C) volume (L) volume reduction factor (–) tangential velocity in the module (m s−1)

Greek Letters

π σ β μ

osmotic pressure (bar) reflection coefficient (–) polarization module (–) viscosity (Pa·s)

show excellent stability, but at the same time produce greater fouling [12]. The surface charge of the membrane depends on the polymeric material used, functional groups, manufacturing method, and feed solution properties [4,12,14–18]. Thin-film composite (TFC) NF membranes contain ionizable groups, such as carboxyl, sulfonic, and amine groups [6,18], which easily dissociate in the presence of acids or bases. Dissociation affects the electrostatic repulsion between ions or charged molecules and the membrane surface. Finally, operating factors such as pressure [5,14,19], feed velocity [5,14,19], temperature [14,19], recovery [7], and continuous filtration operations [9] also influence permeate water transport, fouling, and membrane polarization. These operating factors are usually assessed on a small scale with a flat sheet membrane with an area of 10–100 cm2 [20]. Few investigations have addressed the application of NF in the treatment of mining wastewater [4,5,7,8,15,19–21] by comparing NF with RO. All of the results highlight that NF with circular flat sheet [4,7,15,19–22] and spiral-wound membranes [5,20] is the better option for copper [5,19,21] and gold AMD treatment [7,15,20,22]; however, scaling up these results with flat sheet membranes is a challenge due to the problems by of concentration polarization, fouling, permeability efficiency, and cleaning procedures of spiral-wound modules. In addition, the industrial effluents provide valuable information for scaling up of a separation process, although their use in AMD treatment is limited in the published literature [5,7,8,15,20–22]. Therefore, this study aims to assess the performance of two spiral-wound TFC NF membranes (NF90 and NF270) in the treatment of an actual AMD effluent from an active copper mine. The species removal capacity and permeability of both membranes were investigated by varying the operating conditions such as pressure and feed flow and assaying metal and sulfate. The ability of concentrating AMD treatment to recover water and species was also analyzed with NF270 due to its high treatment capacity and continuous operating performance over a long duration. Finally, membrane recovery and resistance due to scaling after continuous AMD treatment and cleaning procedure were assessed.

metalloids (As, Sb, Se) [2]. In recent years, various technologies have been available for AMD treatment; however, sustainability is deemed the determining factor in selecting an appropriate process [1,2]. Many mining operations use neutralization, in which alkaline agents such as calcium carbonate or lime generate sludge. The treatment of AMD should maximize the recovery of valuable elements such as metals and water in order to compensate for the cost of treatment; however, in neutralization, metal recovery is not economically viable. Membrane separation has become a promising technology for the treatment of contaminated water due to its efficient species removal capacity [3]. Reverse osmosis (RO) and nanofiltration (NF) are the most commonly used membrane separation processes for metal removal [3]. Both of these processes are capable of retaining salts and metals from the feed solution and therefore demonstrate a high potential for the recovery of species and water from AMD [4–7]. However, RO uses membranes that restrict the passage of salt and are therefore less water permeable than NF. In previous studies, Rieger et al. [8] treated a highly concentrated AMD at laboratory scale by NF (NF99) and RO (RO98pHt). In the range of 10 and 40 bar, total ion rejection between 84% and 87% for NF99 and between 90% and 95% for the RO98pHt membrane were obtained, while permeate fluxes were twice as high for NF. During long-term studies, NF also exhibited lower fouling potential. Using the same membranes with AMD at pilot scale, Ambiado et al. [5] similarly observed high permeate fluxes in NF with only slightly lower divalent ion rejection rates. Under optimum pressure conditions of 15 bar, ion removal reached a maximum of 92% for NF99 and 98% for the RO98pHt membrane, while fluxes were respectively 86.2 L/m2 h and 16.5 L/m2 h. Although NF seems to be an attractive alternative for AMD treatment, fouling is still one of the most difficult parameters to control [9]. Fouling depends on the surface properties of the membrane, feed characteristics, and operating conditions [10,11]. The most important studied parameters for NF are roughness, pore size, hydrophobicity, membrane charge, solute concentration in the feed, feed pH, module type, and the hydrodynamic conditions at different transmembrane pressures. Rana and Matsuura [12] discussed the influence of roughness on NF; however, the results of some earlier scientific studies evince that a greater roughness value significantly increases the probability of surface pore blockage generation [4,13]. Membrane hydrophobicity/ hydrophilicity determines its degree of wettability as well as its chemical and mechanical surface stability [14]. Hydrophobic membranes

2. Method and materials 2.1. Membranes Two commercial spiral-wound membranes (NF90 and NF270) manufactured by Dow/Filmtec were used in this experiment. The membranes were selected based on their molecular cut-off, pore size, 115

Chemical Engineering Journal 345 (2018) 114–125

L. Pino et al.

hydrophobicity, roughness, and isoelectric point (IEP). NF90 and NF270 are thin-film composite (TFC) membranes that consist of a microporous polysulfone interlayer and a polyester support layer. The difference between these two TFC membranes is the composition of the active layer. Table 1 shows the properties of each of these membranes.

Table 2 Ionic composition of the studied acid drainage sample. Ion

mg/L

mEq/L

Anions

Sulfate Phosphate Chloride Fluoride Nitrate Boron Silica

2443.50 ± 28.9 0.80 117.30 3.10 42.80 0.02 26.60

50.88 ± 0.49 0.03 3.31 0.16 0.69 – –

Cations

Aluminum Copper Iron Manganese Zinc Calcium Magnesium Potassium Sodium Barium Strontium Ammonium

54.90 ± 0.21 617.90 ± 3.5 2.70 ± 0.01 196.90 ± 0.11 65.80 ± 0.11 117.20 ± 1.24 208.50 ± 2.54 16.80 90.20 0.03 0.80 0.70

6.10 ± 0.02 19.45 ± 0.09 0.15 ± 0.00 7.17 ± 0.00 2.01 ± 0.00 5.85 ± 0.05 17.16 ± 0.17 0.43 3.92 0.00 0.02 0.04

2.2. Mine water composition The AMD used in the study was from an active copper mine in Chile. The original sample was microfiltered in a Rhodia Orelis unit with a ceramic membrane (Kerasep, 0.45 µm of MWCO) in order to remove particles or biological contamination. The compositions of the dissolved species are displayed in Table 2. 2.3. Nanofiltration experimental setup The AMD nanofiltration was performed in an Alfa Laval Pilot Unit (2.5″ RO/NF laboratory unit). It contained a 20-L stainless steel tank, a high-pressure pump, two manometers (0–100 bar) at the inlet and outlet of the membrane, a temperature sensor (0–100 °C), and two flux meters at the permeate and concentrate outlets. A heat exchanger, located after the filtration module, was used to keep the temperature constant with a cryostat (Jeio Tech) with a 50% ethylene glycol/water solution. The pressure and circulation of flow in the pilot unit were adjusted by a manual control valve and a variable-frequency-drive pump. A schematic diagram of the nanofiltration unit is illustrated in Fig. 1.

Ionic imbalance Conductivity (mS/cm) pH

The AMD treatment was performed with a 30-liter solution in total recirculation mode. The filtration was carried out at feed flow rates of 700, 1000, and 1400 L/h at 25 °C and a pressure range of 5–30 bar. The resistance-in-series model was employed to determine the resistance of the membrane to mass transfer. The total resistance is the sum of individual resistances, which include concentration polarization, pore blockage, adsorption, membrane resistance, and scaling due to specific interactions:

2.4. Experimental procedure 2.4.1. Evaluation of the flow rate and pressure The filtration experiments were performed in total recirculation mode to keep the feed concentration constant by returning permeate and concentration flows back to the feed tank. The main purpose of this experiment was to analyze the behavior of permeate flux density, rejection, resistance, and polarization of the membrane in response to changes in pressure and feed flow. First, a distilled water filtration without recirculation was carried out with both membranes to eliminate possible impurities from the system. Next, the system was operated for about 1 h in total recirculation mode at 30 bar and 25 °C until stabilization was achieved. Subsequently, the permeate water flux ( JW ) was measured under controlled conditions (700 L/h and 25 °C) at transmembrane operating pressures (ΔP ) of 4, 8, 10, 20, and 30 bar. The permeate water flux was calculated as:

JW =

ΔV Am ·Δt

6.35% ± 0.24% 6.29 ± 0.01 3.5 ± 0.02

RT =

ΔP−σ·Δπ μ·J

(3)

where Δπ is the osmotic pressure difference, σ is the reflection coefficient, μ is water viscosity at the operating temperature (25 °C), and J is the permeate flux obtained from the AMD treatment. Assuming that fouling includes adsorption resistance, scaling, and pore blockage, total resistance can be defined as the sum of fouling-polarization (RF + P ) and membrane resistance (RM ):

ΔP−σ·Δπ ⎞ ⎛ 1 ⎞ RF + P = RT −RM = ⎜⎛ ⎟−⎜ ⎟ μ·J ⎝ ⎠ ⎝ μ·k w ⎠

(4)

The osmotic pressure difference was calculated based on the proposal of Luo and Wan [10] in order to incorporate the resistance induced by the polarized layer and fouling:

(1)

where ΔV /Δt is the permeate volume over time and Am is the effective filtration area. The hydraulic permeability constant (k w ) was determined using the following expression:

Δπ =

J kw = w ΔP

where Cbi and CPi are the concentrations of solute i in the feed and concentrate, respectively, R is the universal gas constant, and T is the

n



(Cbi−CPi )·R·T

(5)

i=0

(2)

Table 1 Membrane characteristics based on manufacturer and research data. Membrane-module

NF 90–2540

NF 270–2540

Top layer material Cut-off MWCO (g/mol) Area (m2) Rejection-Size (%) Pore diameter (nm) Root mean square roughness (nm) Contact angle (°) Isoelectric point (IEP)

Polyamide, TFC 200 2.6 > 97 (MgSO4) 0.55 [13] 0.68 [23] 0.76 [24] 27.75 [13] 74.9 [25] 40.0 [26] 50.9 ± 4.9 [28] 48.7[25] 42.2 [29] 5.5 [30] 3.9 [28] 4.2 [14]

Polypiperazine amine, TFC 200–300 2.6 > 97 (MgSO4) 0.71 [13] 0.84 [23] 0.88 [24] 4.34 [13] 4.1 [27] 4.0 [26] 28.8 ± 2.4 [28] 32.6 [30] 23.4 [29] 3.0 [30] 2.7 [27] 2.8 [14]

116

Chemical Engineering Journal 345 (2018) 114–125

L. Pino et al.

Fig. 1. Scheme of the nanofiltration pilot unit.

temperature. The reflection coefficient (σ ) was estimated for each solute i [25,31] and then averaged:

(1−Ri )ln(1−Ri ) Ri

σi = 1 +

where v is the tangential velocity in a spiral module, Di is the diffusion coefficient, and dh is the hydraulic diameter. The values of v, Di, dh were calculated using the method proposed by Van Gauwbergen [36] and Andrade et al. [20].

(6)

where Ri is the real (intrinsic) rejection of solute i due to concentration polarization. The reflection coefficient measures the efficiency of the membrane at separating solutes and its values range between 0 and 1, σ = 0 for a nonselective membrane (no solute separation) and σ = 1 for an ideal semi-permeable membrane (no solute transport) [32,33]. Shirazi et al. [33] reported that the solute size becomes equal to or greater than the pore size when σ = 1 and the steric effect is the predominant phenomenon in separation. The disadvantage of the employed model is that it does not include the effects of membrane charge Ri [11]:

Ri = 1−

Cpi Cmi

() ⎡1−exp ( ) ⎤ ⎣ ⎦

R 0i exp = 1−R 0i

2.4.2. Evaluation of feed concentration The effects of feed concentration on permeate flux density, rejection, resistance, and polarization were evaluated in concentration mode. The permeate was collected separately and the rejected component was then recirculated into the feed supply tank until a desired volume reduction factor (VRF ) was attained. VRF can be defined as:

VRF =

(7)

2.4.3. Evaluation of fouling performance The experiment to evaluate fouling was carried out in long-term continuous operation. The system was operated in recirculation mode for a period of ten days at 15 bar and 25 °C with a feed flow rate of 700 L/h. In addition, resistance due to scaling was determined by executing a low-speed physical cleaning with distilled water after the AMD treatment. After completing the cleaning process, hydraulic permeability was determined. Scaling resistance RI can be formulated as:

where Cmi is the solute concentration at the ki membrane surface, ki is the mass transfer coefficient in the polarized layer, and R 0i is the observed rejection. R 0i can be defined as:

R 0i = 1−

Cpi (8)

Cbi

In addition, the polarization module β [31,34] can be calculated as:

β=

Cmi Cbi

(9)

RI = RTF −Rm =

The mass transfer coefficient of each species (ki ) was calculated using Sherwood number (Sh). Koutsou et al. [35] presented a correlation between ki and the spacers of different L/ D ratios (L is the cell length and D is the filament diameter). For the NF270 and NF90 membranes:

Sh = 0.16·Re 0.605 ·Sc 0.42 =

ki·dh Di

ρ ·v ·dh μ

Sc =

μ ρ ·Di

1 1 − kTF ·μ k w·μ

(13)

where RTF and kTF are total membrane resistance and membrane permeability, respectively.

2.4.4. Analyses The conductivity and pH measurements were performed with a YSI conductivity meter (model 3200) and a Hanna Instruments pH meter (model HI 2221), respectively. The ion concentrations of sulfates and metal ions were determined using a spectrophotometer (SulfaVer® 4 kit) and an atomic absorption spectrophotometer (Perkin Elmer, AAnalyst 400), respectively.

(10)

The Reynolds (Re ) and Schmidt (Sc ) numbers are defined as:

Re =

(12)

where VF and VR are the initial feed volume and rejected component volume, respectively.

J ki

J ki

VF VR

(11) 117

Chemical Engineering Journal 345 (2018) 114–125

L. Pino et al.

3. Results and discussion

generated a greater driving force for the diffusion of ions through the membrane. This resulted in a decline in rejection due to the decrease in electrostatic repulsion [32]. Ambiado et al. [5] observed this effect in the NF99 membrane at pressures greater than 20 bar. The reflection coefficient (σ) determines the degree of contributions of convective or diffusive transport in separation. For pressures ranging from 5 to 15 bar, the value of σ varied between 0.74 and 0.86 for NF270, while it was between 0.94 and 0.97 for NF90. For pressures over 15 bar, σ reached averages value of 0.95 and 0.98 for NF270 and NF90, respectively. This means that at low pressures, the convection effect surpassed diffusion, while separation was governed mainly by steric or diffusive rejection at the elevated pressures. The NF270 exhibited significant changes between these two types of transport, whereas the rejection in NF90 was controlled by steric separation. As mentioned in Section 1, the feed pH modifies the NF membrane surface charge. Under the isoelectric point (IEP) of the membrane, the acid-base functional groups are protonated and the membrane charge becomes positive, and vice versa [6]. The membrane charge becomes larger when its distance from the IEP, as well as the density of the functional groups, increases greatly [5,10,18]. Outside the IEP, the coions (ions with the same charge sign as the membrane) are rejected with greater force, while electroneutrality forces the counterions to be as well (Donnan exclusion) [3]. The IEPs of each membrane (found in various investigations) are presented in Table 1, and have an average of ∼2.8 and ∼4.5 for NF270 and NF90, respectively. As the pH of the AMD was 3.5, the surface charge signs of NF270 and NF90 were negative and positive, respectively, and were also dependent on the

The assessment of operating conditions is a useful tool for investigating separation performance and selecting optimal treatment conditions. 3.1. Membrane permeability The hydraulic capacities of the two membranes were determined using pure water permeability, and from Fig. 2 it can be inferred that NF270 membrane showed a greater capacity than NF90. Using Eq. (2), the permeability constants (k w ) of the NF270 and NF90 membranes were found to be 9.05 ± 0.09 and 5.22 ± 0.07 L/m2 h bar, respectively, at 24 ± 1 °C. The obtained k w values were lower than those reported by other authors [4,15,17,22,34] for flat sheet membranes. The Jw results confirmed the characteristics of NF270 and NF90 membrane as loose and tight, respectively [14]. The pore sizes of NF270 and NF90 (Table 1) were consistent with the permeate flux; however, Simon et al. [37] and Artuğ [14] demonstrated similar effective pore radii for both membranes. Therefore, the permeability difference suggests that hydrophilicity is a crucial factor in the AMD treatment. A hydrophilic membrane wets better, thus facilitating flux through the pores. Thus, NF270 could be useful in AMD treatment. 3.2. Influence of applied pressure The AMD permeate flux density ( J ) tests (Fig. 2) determined the critical operating pressure and quantity of permeate water for both the NF270 and NF90 membranes. The behavior of J with increasing pressure coincided with previous investigations on the treatment of effluents using NF with prior microfiltration or ultrafiltration [4,7,8,15,19]. With increasing pressure, J also increased in a linear manner; however, it was observed that the permeate flux density ratio began to decrease after 15–20 bar. The difference between Jw and J under low-pressure conditions (linear zone) occurs due to the osmotic pressure difference between the concentrate and permeate [38]. Under a high operating pressure, Mulder [31] demonstrated that ∂J / ∂ΔP decreased significantly when Δπ was very high, thereby reaching a maximum permeate flux value ( J∞), which depends on the feed concentration (Cb ), concentration at the surface (Cm ), and mass transfer coefficient (ki ). Moreover, the effects of concentration polarization [9], counter-diffusive flux [31], and fouling [33] became important under high pressure. Therefore, the results for both membranes reveal a deviation in permeate flux, however, a maximum permeate flux threshold was not achieved. Schäfer et al. [39] defined the transition point as the critical flux after which membrane fouling becomes significant for long working time intervals. At 15–20 bar, the critical flux for NF270 was found to be 81.13 L/h m2, whereas for NF90 it was 67.10 L/h m2. Ambiado et al. [5] also presented similar results for a spiral-wound NF membrane (NF99). The overall observed rejection (conductivity) and species rejection as a function of pressure for both membranes are displayed in Fig. 3. The results show an increase in rejection as pressure rises. Between 5 and 15 bar, rejections increases were 11.38% and 5.3% for the NF270 and NF90 membranes, respectively, while above 15 bar, they were 6.67% and 0.5% respectively. Thus, NF270 exhibited significant changes on rejection with increasing pressure compared to NF90. The results regarding rejection depend on permeate conductivity, which in turn depends on the relative flux of water and ions through the membrane. Pressure boosted the flux of water through the membrane, but the transport of metals and sulfate was obstructed by the steric effect and surface charge effects (electrostatic repulsion) [32,40], thus resulting in a greater species dilution in the permeate and an increase in rejection due to high solvent flux (Eq. (8)). Further, when the pressure reached even higher, the ion transport toward the membrane increased, and hence, the concentration polarization became conspicuous as it

Fig. 2. Effect of pressure and feed flow on membrane permeability with AMD in NF270 (A) and NF90 (B) at 25 ± 1 °C. 118

Chemical Engineering Journal 345 (2018) 114–125

L. Pino et al.

order of distribution is the inverse of the obtained order of species rejection (it occurs only if the single divalent ion fraction is considered). The species fraction associated with sulfate could produce larger compounds and cause an increase in species rejection. Rejection (conductivity) exhibited a significant difference compared to the individual species rejection in both membranes (Fig. 3). Childress and Elimelech [18] explained that the high conductivity associated with H+ protons (3.5·10−2 S m2 /mol ) in the permeate significantly controlled the overall rejection, unlike the cases of copper (1.07·10−2 S m2 /mol ), aluminum (1.83·10−2 S m2 /mol ), zinc (1.05·10−2 S m2 /mol ), or sulfate ions (1.6·10−2 S m2 /mol ) [44]. The increase in H+ concentration in permeate flux with NF has been demonstrated in earlier works [5,10]. Therefore, it can be inferred that the overall species rejection was lower than the individual species rejections due to the increase of monovalent ions, such as H+, toward the permeate. On average, the difference between the global rejection and rejections by ion were 14.26% and 6.41% for NF270 and NF90, respectively. The effect of the change in operating pressure on resistance due to membrane fouling-polarization is displayed in Fig. 4. RF + P remained practically constant at the pressures below 15 bar and then increased almost linearly. The NF270 membrane was more prone to fouling and polarization during treatment but had a greater permeate flux; total AMD filtration resistance was lower for NF270. RF + P was also consistent with the surface properties of the membrane (Table 1) as NF270 generated less resistance to water permeability due to lower surface roughness a lower contact angle [12]. 3.3. Influence of feed velocity or feed flow As feed velocity has an impact on concentration polarization, the changes in tangential velocity in the spiral-wound module were studied. In addition, the effects of feed velocity on permeate flux density, rejection, and resistances were also explored. The feed flows were chosen from the normal operating range of an NF plant with spiral-wound modules. The Reynolds number did not exceed the transition regime and oscillated between 100 and 1000 [34]. The assessed feed flows were 700, 1000, and 1400 L/h , with tangential velocities of 0.209, 0.298 and 0.417 m/s and Reynolds numbers of 162, 231, and 324 , respectively. For the studied Re range in both membranes, there was no discernable effect on permeate flux density as a result of the change in feed velocity from a low to high value (Fig. 2). Ambiado et al. [5] demonstrated similar findings; however, according to Rieger et al. [8] and AlZoubi et al. [19], the increase in feed flow in flat sheet membranes caused an increase in permeate flux density. As these studies did not report velocities or Re, it is impossible to establish any comparison. Andrade et al. [20] explained the effect of the change in Re (from 8 to 15) in low-velocity ranges. Fig. 5 shows a slight increase in rejection (conductivity) with an increase in feed velocity for both membranes. It is evident that the real rejections (R ) were higher than the observed rejections (Ro ) and the difference decreased as Re increased. This behavior can be explained by the decrease in membrane polarization due to the increase in feed velocity and shear strain [40]. In addition, the change in velocity also increases the mass transfer coefficient (k ) of the

Fig. 3. Effects of pressure on conductivity and observed sulfate, copper, aluminum and zinc rejection at a feed flow of 1000 L/h and 25 ± 1 °C for NF270 (A) and NF90 (B).

quantity of carboxyl and amine groups of the active layer or TFC. The surface charge of NF90 generated greater electrostatic cation or metal repulsion than NF270 (Fig. 3); however, at pressures above 25 bar, the difference in rejection between the two membranes was minimal. A high metal rejection in NF90 also caused a high sulfate rejection to maintain electroneutrality (there was 50.88 mEq/l of the sulfate anion in the AMD). However, sulfate rejection by NF270 was not high in spite of its negative charge. Therefore, it can be posited that the low electrostatic rejection by NF270 occurred due to the negative surface charge (zeta potential) resulting from its proximity to the IEP [6] and the permeation of cations through the membrane [17]. Finally, the charge difference between the two membranes was trivial at higher pressures (> 25 bar) because of the steric effect, a predominant phenomenon in separation using membranes with small pores [41]. Metal rejection was highest for Al and lowest for Cu (Al > Zn > Cu) in both membranes. This behavior was consistent with the increase in standard molar Gibbs energy of solvation or hydration of ions [11,32] (Table 3). Cations and anions in solutions are surrounded by water molecules; the ions with the smallest radius and highest charge generate the highest hydration energy [42], leading to an increase in the hydrated radius. Therefore, higher rejection of Al ions was obtained compared to that of Cu and Zn ions by steric separation. Moreover, the chemical speciation analysis of Al, Zn, Cu, and sulfate at a pH of 3.5 indicates the partial formation of other charged and uncharged species [43]. For aluminum, the species fractions of AlSO4+, Al (SO4 )−2 , and Al+3 were 79.9%, 17.57%, and 3.6%, respectively; for zinc, the species fractions of ZnSO4 , Zn+2 , and Zn (SO4 )−2 2 were 70.1%, 23.9% and 6%, respectively; and finally, for copper, the species fractions of CuSO4 and Cu+2 were 71.8%, and 28.2%, respectively. This

Table 3 Radius (r), width of hydration shell (Δr), number of water molecules in the shell (n) and standard molar Gibbs energy of hydration of ions (ΔhydG○) at 298.15 K [45].

119

Ion

r [nm]

Δr [nm]

n

ΔhydG° [kJ mol−1]

Aluminum Zinc Copper Sulfate

0.053 0.075 0.096 0.230

0.324 0.220 0.224 0.043

20.4 9.6 9.9 3.1

−4525 −1955 −2010 −1080

Chemical Engineering Journal 345 (2018) 114–125

L. Pino et al.

Fig. 4. Variation of total resistance with pressure for NF270 (A) and NF90 (B). For comparison, membrane resistance Rm is also shown.

mentioning that the effluents previously underwent ultrafiltration; therefore, the increase in resistance could not be explained by particle accumulation. Moreover, oversaturation of solids that cause scaling in AMD NF, such as CaSO4 [46], and the increase in turbidity in total recirculation mode were not detected.

polarized layer due to the increase in Sh number (Eq. (10)). The decrease in polarization was more detectable in NF270 than NF90 because NF270 generated greater polarization due to higher permeate flux. It was also noticed that the increase in velocity did not produce any significant increase in rejection (real or observed) in either membrane. The effect of feed velocity change on the β module is presented in Fig. 6. It is evident that β values were consistent with the typical concentration polarization of inorganic effluents (< 2) [46]. The increase in Re value from 162 to 324 caused a decrease in concentration polarization (β ) by approximately 34.7% and 30.9% in NF270 and NF90, respectively, and consequently, a slight increase in species rejection (Fig. 5). Moreover, the rise in the osmotic pressure of the solution resulted in the decrease in ∂J / ∂ΔP . However, the increase in velocity or Re did not affect the permeate flux (Fig. 2). Fig. 4 depicts the effect of the change in Re on total resistance at different pressures. The hydrodynamics of tangential filtration systems, especially the increase in feed velocity, have a strong influence on total resistance or resistance due to fouling and polarization [33]. The increase in velocity caused an increase in shear strain on the surface, thus limiting the accumulation of particles and ions on it; however, the effect depends on several factors such as the nature of the particles and ions and the pressure applied to the system [33]. The obtained results did not have a clear relationship to the increase in velocity, which could be due to because polarization layer in the employed velocity was range not being significantly affected by the change in Re. It is worth

3.4. Feed concentration influence Volume reduction tests were carried out to concentrate the AMD. A pressure of 20 bar and with a feed flow of 700 L/h were selected in order to have a low resistance and high permeate flux by eliminating concentration polarization and minimizing energy consumption. The study was executed with NF270 due to its high treatment capacity, acceptable rejection, low resistance to filtration, and low polarization at moderate pressures. At each VRF, various samples were taken to determine sulfate and metal contents. The rejections (real and observed) and reflection coefficient were calculated on the basis of conductivity using Eqs. (6)–(8). The osmotic pressure was calculated with the sum of the initial concentrations of the analyzed species at different VRF assuming an associated error of ∼10%. Fig. 7 displays the effect of the increase in VRF on permeate flux density and observed rejection, while the results of the species, osmotic pressure, reflection coefficient, and total resistance analyses are shown in Table 4. It is noticeable that the decrease in flux density coincided with the increase in total resistance due to the increase in feed 120

Chemical Engineering Journal 345 (2018) 114–125

L. Pino et al.

Fig. 5. Effect of Re number on overall rejection by NF270 (A) and NF90 (B) (observed rejection – solid symbols, real rejections – empty symbols).

thus producing a change in effective volumetric charge [40]. The results regarding species concentration in the concentrate (Table 5 and Fig. 9) show that the increase in Ci/ Ci0 was proportional to the increase in VRF and that metals exhibited a more significant increase. The lack of correlation was due to the loss of ions the permeate flux and the accumulation of ions on the membrane. It is worth noting that Cu exhibited the smallest loss, thus promoting its recovery in the concentrate. According to Al-Zoubi et al. [21], the decrease in sulfate was due to the formation of scaling such as CaSO4 [46] in the concentrate. Finally, the permeate flux density and membrane rejection analyses based on the VRF measurements provide important information for scaling up in continuous AMD treatment with variable concentrations. While the permeate flux decreased by about 50%, rejection remained high without affecting the permeate quality even though the feed conductivity was increased elevenfold (Fig. 8).

concentration. The fouling-polarization component of total membrane resistance increased from 43.93% (VRF = 1) to 65.50% (VRF = 4). In addition, the decrease in effective pressure (ΔP−σ Δπ ) due to the rise in osmotic pressure also contributed to the decrease in permeate flux density (Fig. 8). Therefore, it can be inferred that the increase in concentration caused an increase in species deposition or scaling on the surface. Al-Rashdi et al. [4] showed that copper induced blockages and fouling on the surface and its adsorption increased when the solution had a low pH. In addition, permeate flux density (Fig. 7) exhibited an initial period of rapid flux decline followed by a small change until steady state was reached. It is noticed that there were an increase and slight decrease in the observed rejection due to the increase in VRF (Fig. 7). The reason for this behavior is still unclear; however, Fang and Deng [47] obtained similar results in the treatment of As (V) with NF membranes. The behavior could be explained by the attraction or union of species on the membrane, resulting in a decrease in effective pore size [42,48]. Tansel et al. [42] reported that larger species can easily lose their hydrated layer when they lie near the membrane, thus generating a stronger bond with the surface. The reflection coefficient (σ ) results indicate that there was a decrease in convective transport due to the decrease in effective pore size [33]. In addition to the size exclusion effect, rejection in NF can also be affected by an inversion of the membrane surface charge [48] caused by the increase in concentration. This phenomenon occurs when the membranes have a negative surface charge and predominantly attract cations in the vicinity of the active membrane layer,

3.5. Long-term The performance of NF270 in response to exposure time was assessed by evaluating the permeate flux and rejection based on conductivity. In Fig. 10A, it is noticed that permeate flux density decreased for at least seven days and then remained constant. This phenomenon can be explained by the temporal evolution of RF + P , in which after an initial rise, RF + P reached a steady state in response to filtration (Fig. 10B). The RF + P increased by 39.8% to 46.7%, it explains the 121

Chemical Engineering Journal 345 (2018) 114–125

L. Pino et al.

Fig. 6. Effect of Re number on the polarization modulus (β) for NF270 (A) and NF90 (B).

decline of permeate flux ∼10%. In addition, the decline due to the fouling followed a particular pattern that was similar to the experiment carried out by Lin et al. [49]. In their research, a four-stage permeate flux decline model (polarization, nucleation, crystallization and deposition decline, and steady state) was proposed for CaSO4 fouling in an

NF membrane. In order to determine scaling or species precipitation fouling, a physical cleaning at low velocity followed by a hydraulic permeability characterization was carried out. It was found that resistance due to scaling and polarization were 1.74 × 1013 [m−1] and 2.05 × 1013 [m−1],

Fig. 7. Normalized permeate flux and real rejection of the NF270 membrane at 20 bar, 700 L/h and 25 °C as function of VRF. 122

Chemical Engineering Journal 345 (2018) 114–125

L. Pino et al.

Table 4 Compositions of the different VRF solutions. VRF

π [bar]

Feed [mg/L] Sulfate

1 2 3 4

2613.0 3411.0 4913.0 7637.0

± ± ± ±

10.0 20.5 15.0 22.5

Copper

Aluminum

Zinc

673.1 ± 22.1 1292.8 ± 25.1 1838.5 ± 31.1 2512.2 ± 43.9

302.9 ± 8.2 547.4 ± 16.1 788.8 ± 5.0 1066.5 ± 3.7

69.15 ± 1.2 113.17 ± 1.0 164.50 ± 1.3 232.83 ± 1.4

1.26 1.96 2.82 4.09

± ± ± ±

0.02 0.03 0.02 0.02

σ

RT [×1013 m−1]

0.712 0.789 0.786 0.781

7.91 11.6 12.0 12.9

± ± ± ±

0.01 0.01 0.01 0.01

Fig. 8. Variation of the effective pressure and conductivity of the permeate and feed for the NF270 membrane at 20 bar, 700 L/h and 25 °C as a function of VRF.

respectively. Therefore, the contribution of polarization and scaling to fouling was 22.81% and 24.30%, respectively, when the steady state was reached. Moreover, the observed rejections yielded a high removal of metals and sulfate. The increase in rejection (∼7%) at the beginning of the treatment was not completely discernible. The decrease in pore radius (blockage), charge inversion, or simply an adjustment dynamic in the active membrane layer could be the reasons for this behavior. The results of the cleaning process regarding permeability and total resistance are presented in Table 5. The decline of permeability in NF270 was close to 28% after the AMD treatment. In addition, a good membrane recovery (99%) was observed after the alkaline and acid treatment. Therefore, the membrane recovery results indicate that the membrane is applicable in the long-term continuous operation process.

Table 5 Hydraulic permeability and total resistance of NF270 before and after cleaning process. Condition

Hydraulic permeability [L/ m2 h bar]

Total resistance [×1013 m−1]

Before treatment of AMD After treatment of AMD After cleaning

9.01 ± 0.115

4.45 ± 0.057

5.67 ± 0.038

7.08 ± 0.047

8.98 ± 0.161

4.47 ± 0.080

Fig. 9. Variation of the species concentration ratio in the concentrate for the NF270 membrane at 20 bar, 700 L/h, and 25 °C as a function of VRF. 123

Chemical Engineering Journal 345 (2018) 114–125

L. Pino et al.

Fig. 10. Variation of (A) permeate flux and rejection, (B) fouling-polarization resistance over time of NF270 at 700 L/h and 10 bar at 25 ± 1 °C.

4. Conclusions

Acknowledgements

The treatment of industrial effluents is generally difficult due to their chemical variability. Nonetheless, commercial spiral-wound membranes demonstrated a high metal and sulfate removal capacity during both concentration and continuous AMD operations. The increase in operating pressure had a strong influence on permeate flux in both membranes, it was observed that the difference in permeate flux between the NF270 and NF90 membranes noticeably increased at pressures greater than 15 bar. Furthermore, in comparison to NF90, NF270 demonstrated a higher initial rejection increase rate with pressure; however, no difference in species removal between these two membranes was noticed when the pressure exceeded 25 bar. The increase in Reynolds number had no discernible effect on permeate flux or species rejection, but as the pressure increased, the change in Re (low to high) caused a decrease in the polarization module (β) in both membranes. The low resistance to filtration, high removal capacity, low concentration polarization, and high permeate flux at low operating pressures were the main factors for selecting the NF270 for the concentration and continuous operation tests. The rise in feed concentration caused a decrease in permeate flux and an increase in rejection due to the increase in resistance and decrease in effective pressure. Continuous operation showed that resistance produced by fouling and polarization accounted for 46.7% of the total resistance and also decreased the permeate flux by ∼10%. In addition, an excellent membrane recovery (99%) was observed after the alkaline and acid cleaning.

We would like to acknowledge the financial support to this study by FONDAP/CRHIAM Project 15130015 and CONICYT-PCHA/National Doctorate 2016 - 21160607 Chile. References [1] G.S. Simate, S. Ndlovu, Acid mine drainage: challenges and opportunities, J. Environ. Chem. Eng. 2 (2014) 1785–1803. [2] B. Vital, J. Bartacek, J.C. Ortega-Bravo, D. Jeison, Treatment of acid mine drainage by forward osmosis: heavy metal rejection and reverse flux of draw solution constituents, Chem. Eng. J. 332 (2018) 85–91. [3] S. Zhao, C. Ba, Y. Yao, W. Zheng, J. Economy, P. Wang, Removal of antibiotics using polyethylenimine cross-linked nanofiltration membranes: relating membrane performance to surface charge characteristics, Chem. Eng. J. 335 (2018) 101–109. [4] B.A.M. Al-Rashdi, D.J. Johnson, N. Hilal, Removal of heavy metal ions by nanofiltration, Desalination 315 (2013) 2–17. [5] K. Ambiado, C. Bustos, A. Schwarz, R. Borquez, Membrane technology applied to acid mine drainage from copper mining, Water Sci. Technol. 75 (2016) 705–715. [6] M. Mullett, R. Fornarelli, D. Ralph, Nanofiltration of mine water: impact of feed pH and membrane charge on resource recovery and water discharge, Membranes 4 (2014) 163–180. [7] B.C. Ricci, C.D. Ferreira, A.O. Aguiar, M.C.S. Amaral, Integration of nanofiltration and reverse osmosis for metal separation and sulfuric acid recovery from gold mining effluent, Sep. Purif. Technol. 154 (2015) 11–21. [8] A. Rieger, P. Steinberger, W. Pelz, R. Haseneder, G. Härtel, Mine water treatment by membrane filtration processes – experimental investigations on applicability, Desalin. Water Treat. 6 (2009) 54–60. [9] S.S. Bucs, N. Farhat, J.C. Kruithof, C. Picioreanu, M.C.M. van Loosdrecht, J.S. Vrouwenvelder, Review on strategies for biofouling mitigation in spiral wound membrane systems, Desalination (2018). [10] J. Luo, Y. Wan, Effects of pH and salt on nanofiltration—a critical review, J. Membr. Sci. 438 (2013) 18–28. [11] C.-V. Gherasim, P. Mikulášek, Influence of operating variables on the removal of

124

Chemical Engineering Journal 345 (2018) 114–125

L. Pino et al.

[12] [13] [14] [15]

[16]

[17]

[18]

[19]

[20]

[21] [22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

heavy metal ions from aqueous solutions by nanofiltration, Desalination 343 (2014) 67–74. D. Rana, T. Matsuura, Surface modifications for antifouling membranes, Chem. Rev. 110 (2010) 2448–2471. N. Hilal, H. Al-Zoubi, N.A. Darwish, A.W. Mohammad, Characterisation of nanofiltration membranes using atomic force microscopy, Desalination (2005) 187–199. G. Artug, Modelling and Simulation of Nanofiltration Membranes, Cuvillier, 2007. A.O. Aguiar, L.H. Andrade, B.C. Ricci, W.L. Pires, G.A. Miranda, M.C.S. Amaral, Gold acid mine drainage treatment by membrane separation processes: an evaluation of the main operational conditions, Sep. Purif. Technol. 170 (2016) 360–369. A.W. Mohammad, Y.H. Teow, W.L. Ang, Y.T. Chung, D.L. Oatley-Radcliffe, N. Hilal, Nanofiltration membranes review: recent advances and future prospects, Desalination 356 (2015) 226–254. C. Bellona, J.E. Drewes, The role of membrane surface charge and solute physicochemical properties in the rejection of organic acids by NF membranes, J. Membr. Sci. 249 (2005) 227–234. A.E. Childress, M. Elimelech, Relating nanofiltration membrane performance to membrane charge (electrokinetic) characteristics, Environ. Sci. Technol. 34 (2000) 3710–3716. H. Al-Zoubi, A. Rieger, P. Steinberger, W. Pelz, R. Haseneder, G. Härtel, Optimization study for treatment of acid mine drainage using membrane technology, Sep. Sci. Technol. 45 (2010) 2004–2016. L.H. Andrade, A.O. Aguiar, W.L. Pires, L.B. Grossi, M.C.S. Amaral, Comprehensive bench- and pilot-scale investigation of NF for gold mining effluent treatment: membrane performance and fouling control strategies, Sep. Purif. Technol. 174 (2017) 44–56. H. Al-Zoubi, A. Rieger, P. Steinberger, W. Pelz, R. Haseneder, G. Härtel, Nanofiltration of acid mine drainage, Desalin. Water Treat. 21 (2010) 148–161. L.H. Andrade, B.C. Ricci, L.B. Grossi, W.L. Pires, A.O. Aguiar, M.C.S. Amaral, Nanofiltration applied in gold mining effluent treatment: evaluation of chemical cleaning and membrane stability, Chem. Eng. J. 323 (2017) 545–556. L.D. Nghiem, S. Hawkes, Effects of membrane fouling on the nanofiltration of pharmaceutically active compounds (PhACs): mechanisms and role of membrane pore size, Sep. Purif. Technol. 57 (2007) 176–184. M.J. López-Muñoz, A. Sotto, J.M. Arsuaga, B. Van der Bruggen, Influence of membrane, solute and solution properties on the retention of phenolic compounds in aqueous solution by nanofiltration membranes, Sep. Purif. Technol. 66 (2009) 194–201. Y.L.G. De Schwarz, D.A. Cornwell, Y.A.L. Gouellec, A.R. Foundation, L.B.W. Department, R.C. Cheng, A Novel Approach to Seawater Desalination Using Dual-staged Nanofiltration, Awwa Research Foundation, 2006. A. Subramani, X. Huang, E.M. Hoek, Direct observation of bacterial deposition onto clean and organic-fouled polyamide membranes, J. Colloid Interface Sci. 336 (2009) 13–20. K.L. Tu, A.R. Chivas, L.D. Nghiem, Effects of membrane fouling and scaling on boron rejection by nanofiltration and reverse osmosis membranes, Desalination 279 (2011) 269–277. K.L. Tu, L.D. Nghiem, A.R. Chivas, Coupling effects of feed solution pH and ionic strength on the rejection of boron by NF/RO membranes, Chem. Eng. J. 168 (2011) 700–706. L.D. Nghiem, D. Vogel, S. Khan, Characterising humic acid fouling of nanofiltration

[30]

[31] [32]

[33] [34] [35]

[36] [37]

[38] [39] [40]

[41]

[42]

[43] [44] [45] [46] [47]

[48]

[49]

125

membranes using bisphenol A as a molecular indicator, Water Res. 42 (2008) 4049–4058. V.T. Do, C.Y. Tang, M. Reinhard, J.O. Leckie, Degradation of polyamide nanofiltration and reverse osmosis membranes by hypochlorite, Environ. Sci. Technol. 46 (2012) 852–859. M. Mulder, Basic Principles of Membrane Technology, Springer, 1996. C.-V. Gherasim, J. Cuhorka, P. Mikulášek, Analysis of lead(II) retention from single salt and binary aqueous solutions by a polyamide nanofiltration membrane: experimental results and modelling, J. Membr. Sci. 436 (2013) 132–144. S. Shirazi, C.-J. Lin, D. Chen, Inorganic fouling of pressure-driven membrane processes – a critical review, Desalination 250 (2010) 236–248. A.J.C. Semião, A.I. Schäfer, Estrogenic micropollutant adsorption dynamics onto nanofiltration membranes, J. Membr. Sci. 381 (2011) 132–141. C.P. Koutsou, S.G. Yiantsios, A.J. Karabelas, A numerical and experimental study of mass transfer in spacer-filled channels: effects of spacer geometrical characteristics and Schmidt number, J. Membr. Sci. 326 (2009) 234–251. D. Van Gauwbergen, J. Baeyens, Macroscopic fluid flow conditions in spiral-wound membrane elements, Desalination 110 (1997) 287–299. A. Simon, J.A. McDonald, S.J. Khan, W.E. Price, L.D. Nghiem, Effects of caustic cleaning on pore size of nanofiltration membranes and their rejection of trace organic chemicals, J. Membr. Sci. 447 (2013) 153–162. M.D. Afonso, M.N. de Pinho, Transport of MgSO4, MgCl2, and Na2SO4 across an amphoteric nanofiltration membrane, J. Membr. Sci. 179 (2000) 137–154. A. Schäfer, N. Andritsos, A.J. Karabelas, E. Hoek, R. Schneider, M. Nyström, Fouling in Nanofiltration, Elsevier, 2004. C. Labbez, P. Fievet, A. Szymczyk, A. Vidonne, A. Foissy, J. Pagetti, Retention of mineral salts by a polyamide nanofiltration membrane, Sep. Purif. Technol. 30 (2003) 47–55. J. Schaep, C. Vandecasteele, A.W. Mohamma, R. Bowen, Modelling the retention of ionic components for different nanofiltration membranes, Sep. Purif. Technol. 22–23 (2001) 169–179. B. Tansel, J. Sager, T. Rector, J. Garland, R.F. Strayer, L. Levine, M. Roberts, M. Hummerick, J. Bauer, Significance of hydrated radius and hydration shells on ionic permeability during nanofiltration in dead end and cross flow modes, Sep. Purif. Technol. 51 (2006) 40–47. I. Puigdomenech, Hydra/Medusa Chemical Equilibrium Database and Plotting Software, KTH Royal Institute of Technology, 2004. D. Electrolyte, Conductance measurements Part 1: theory, Curr. Sep. 18 (1999) 92. Y. Marcus, Thermodynamics of solvation of ions. Part 5. – Gibbs free energy of hydration at 298.15 K, J. Chem. Soc. Faraday Trans. 87 (1991) 2995–2999. W. Guo, H.H. Ngo, J. Li, A mini-review on membrane fouling, Bioresour. Technol. 122 (2012) 27–34. J. Fang, B. Deng, Rejection and modeling of arsenate by nanofiltration: contributions of convection, diffusion and electromigration to arsenic transport, J. Membr. Sci. 453 (2014) 42–51. C.V. Chung, N.Q. Buu, N.H. Chau, Influence of surface charge and solution pH on the performance characteristics of a nanofiltration membrane, Sci. Technol. Adv. Mater. 6 (2005) 246–250. C.-J. Lin, S. Shirazi, P.S. Shakya, Closure to “Mechanistic model for CaSO4 fouling on nanofiltration membrane” by Che-Jen Lin, Saqib Shirazi, and Pritesh Rao, J. Environ. Eng. 133 (2007) 942–943.