Nanofiltration of bulk drug industrial effluent using indigenously developed functionalized polyamide membrane

Chemical Engineering Journal 233 (2013) 193–200

Contents lists available at ScienceDirect

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

Nanofiltration of bulk drug industrial effluent using indigenously developed functionalized polyamide membrane B. Venkata Swamy a, M. Madhumala b, R.S. Prakasham c, S. Sridhar b,⇑ a

Department of Biotechnology, Padmasri Dr. B.V. Raju Institute of Technology, Narsapur, Medak 502313, AP, India Membrane Separations Group, Chemical Engineering Division, Indian Institute of Chemical Technology, Hyderabad 500007, AP, India c BEEC Division, Indian Institute of Chemical Technology, Hyderabad 500007, AP, India b

h i g h l i g h t s  Development of 4-layered indigenous functionalized nanofiltration membrane.  Processing of bulk drug industrial effluent by nanofiltration.  Development of mathematical model and simulation program using statistical mechanical equations.  Scale-up and economic estimation for comparison of NF and RO systems.

a r t i c l e

i n f o

Article history: Received 16 April 2013 Received in revised form 9 August 2013 Accepted 13 August 2013 Available online 21 August 2013 Keywords: Bulk drug industrial effluent Nanofiltration Reverse osmosis Mathematical modeling Economic estimation

a b s t r a c t Nanofiltration (NF) and reverse osmosis (RO) and are membrane-based separation processes which can be used for treatment of industrial effluents and wastewater recycling. In the present study, performance of functionalized nanofiltration (FNF-400) and thin film composite (TFC) polyamide RO membranes has been investigated for the treatment of industrial effluent consisting of 5710 ppm total dissolved solids (TDS), 4050 ppm chemical oxygen demand (COD) and 8.4 ppm biochemical oxygen demand (BOD). Effect of various parameters such as applied pressure and feed composition on parameters such as permeate conductivity and flux, TDS rejection, turbidity removal besides COD and BOD reduction was evaluated. At a constant feed pressure of 21 bar, higher average flux of 36.95 L/m2 h was observed in case of NF when compared to 18.77 L/m2 h for RO. The % rejections of TDS, turbidity and COD were observed to be 85%, 97.8% and 73.33% for NF and 95%, 100% and 86.66%, respectively for RO systems. A statistical mechanical model was developed for commercial NF and RO systems. The economic estimation of commercial NF/RO systems was carried out which reveals that NF process is more economical due to lower operating pressures and consequently lesser energy consumption. Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction High pressure membrane processes such as NF and RO are increasingly used for desalination of seawater, especially in coastal regions to combat high salinity in brackish water and water scarcity. In many regions ground water quality is one of the important factors for which NF/RO membranes have been applied for the production of safe and clean drinking water at low operating pressure with high water flux [1]. There is a growing awareness on the importance of trace levels of contaminants such as endocrine disrupting compounds (EDCs), pharmaceutically active compounds (PhACs), fluorides and dyes originating from industrial, agricultural, medical and domestic processes [2,3]. Compounds used in pharmaceuticals, personal care products and other consumables ⇑ Corresponding author. Tel.: +91 40 27193408/27191394; fax: +91 40 27193626. E-mail address: [email protected] (S. Sridhar). 1385-8947/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cej.2013.08.045

may enter aquatic environments after passing though wastewater treatment plants (WWTP), which are often not designed to remove such chemicals [4–6]. An integrated understanding of organic micro pollutant rejection mechanisms by NF/RO membranes has put the perspective on solutemembrane interactions which are influenced by process conditions and feed water composition [7,8]. A key component in a successful sustainable and economical implementation of NF/RO systems in wastewater polishing is to find a disposal or reuse strategy for the resulting concentrates/ rejects [9,10]. Concentrates from NF process cannot be discharged directly into water bodies due to legislation and environmental regulations. For any discharge concept, apart from dilution to meet discharge levels, it is imperative to minimize the concentrate volume [11]. Minimization of the concentrate is often hampered by severe fouling effects with the main fouling constituents in the effluent being biologically inert dissolved organics originating from the WWTP influent and biological residue products from the

194

B. Venkata Swamy et al. / Chemical Engineering Journal 233 (2013) 193–200

Nomenclature Aw Cp Cf CW Cb DP JV JW Ui

rT li Fi DP DPm Dij DiM BO H

aI

DTij

water permeability co-efficient (m3 m/m2 s mm Hg) concentration of the permeate (mol/m3) concentration of the feed (mol/m3) membrane wall concentration (mol/m3) bulk solution concentration (mol/m3) osmotic pressure (bar) volumetric flux (L/m2 h) water flux (L/m2 h) transport velocity of species i and j (m/s) equivalent isothermal gradient of chemical potential of species i external force per mole on species i total pressure gradient (bar) trans membrane pressure (bar) diffusion coefficient between species i and j within the membrane (m2/s) diffusion coefficient between species i and the membrane (m2/s) viscous flow parameter viscosity of the solution in the membrane (kg/m s) Dimensionless parameter multi-component thermal diffusion coefficient (m2/s)

activated sludge process [12]. To overcome these problems, several researchers have recently focused their work on the removal of organic micro pollutants, EDCs, PhACs, fluoride and dyes using NF/RO membrane technology [13]. Amjad [14] tested membrane-based separation processes like ultrafiltration (UF) and reverse osmosis (RO) for treating a wide variety of industrial effluents. Jain et al. [15] used NF and RO processes for the treatment of leather plant effluent. From their studies, BOD and COD values of treated effluent were found to be well within the permissible limits. Tinghui et al. [16] reported rejection values of P90% for most types of ionic compounds by RO membranes. Erswell et al. [17] used NF membranes to retain organic compounds of low molecular weight such as hydrolyzed reactive dyes as well as dyeing auxiliaries, divalent ions and certain large monovalent ions. In the present investigation, an attempt was made to compare the efficiency of NF and RO processes for the treatment of bulk drug industrial effluent. Color, TDS, conductivity, turbidity, COD and BOD contents present in the feed, permeate and reject samples were analyzed by standard methods. The effect of various operating conditions on water flux and % rejection has been evaluated. With the obtained results, a mathematical model based on statistical mechanical transport equations was developed for commercial RO system. A detailed economic estimation of commercial NF/RO systems is presented.

2. Experimental work

LCij Lo

a1 ; a2 Lp

X

rv rs Pe P Dz C1 C2 D1 D2 R T ni Ci

Onsager coefficient in a center of mass frame of reference viscous flow coefficient coefficients that describe viscous separation effects hydraulic conductivity (m/s) coefficient of solute permeability reflection coefficients for volume flow reflection coefficient for solute flux Peclet number solute diffusive permeability (m/s) effective thickness of the membrane (mm) diffusion factor selectivity factor flow factor (mPa1 s1) membrane constant (kg m2 Pa1 s1) universal gas constant (8.314 J K1 mol1) absolute temperature (301 K). the number of ions formed when the solute dissociates the TDS concentration (mol L1)

Winkler’s reagent, MnSO4, potassium iodide, starch indicator, citric acid, HCl, EDTA, NaOH, hexane and sodium metabisulphite (SMBS) were purchased from sd Fine Chemicals Ltd., Hyderabad, India. Trimesoyl chloride, piperazine and meta-phenylenediamine were obtained from Sigma Aldrich, Bangalore, India. BOD incubator (RCI-S.NO-313, India), COD analyzer (DRB 200 COD Reactor, Germany) for determination of biochemical and chemical oxygen demand besides Colorimeter (Hach-DR-890) for turbidity analysis were procured from Mns Hach, Bangalore, India. Conductivity meter (DCM-900) and pH meter (DPH-504) were purchased from Global Electronics, Hyderabad, India. 2.2. Synthesis of functionalized NF and TFC polyamide RO membranes Polyethersulfone (PES) ultraporous substrate of approximately 50 kDa molecular weight cutoff (MWCO) was prepared by phase inversion method using 15% w/v solution of the polymer in dimethyl formamide (DMF) solvent containing 3% v/v propionic acid [18,19]. The homogenous bubble free solution was then cast on a nonwoven polyester fabric support affixed onto a clean glass plate using a doctor’s blade followed by immersion in ice cold water bath to obtain ultraporous substrate. To obtain polyamide NF membrane by interfacial polymerization, PES substrate was soaked in 1% aqueous solution of piperazine for 1 min. After draining off excess water, the substrate was immersed in hexane bath containing 0.1% of trimesoyl chloride (TMC) for 30 s. The membrane was then heated in an oven at 110 °C for 10 min to obtain a NF membrane of about 600 MWCO. 0.75% dilute polyvinylalcohol (PVA)

2.1. Materials Indigenously synthesized membranes of TFC polyamide RO and functionalized polyamide (FNF-400) were scaled-up to spiral wound membrane modules having effective separation area of 2.5 m2 each, with the help of facilities available with Permionics Membranes Pvt. Ltd., Vadodara, India. The bulk drug industrial effluent was obtained from GVK Bio Pharma Ltd., IDA Nacharam, Hyderabad, India and the respective characteristics are depicted in Table 1. Potassium dichomate, ferrous ammonium sulfate, mercuric sulfate, sulfuric acid, ferroin indicator, sodium thiosufate,

Table 1 Feed characteristics of bulk drug industrial effluent. Feed characteristics TDS (ppm) Conductivity (mS/cm) Turbidity (FAU) Color (Pt–Co) pH BOD (ppm) COD (ppm)

5710 8.45 318 60 6–6.5 8.409 4050

B. Venkata Swamy et al. / Chemical Engineering Journal 233 (2013) 193–200

195

Fig. 1. (a) Process Flow diagram for preparation of FNF membrane (b) Physical structure of FNF-250 membrane, (c) Sketch of spiral wound module for NF/RO process.

solution was prepared by dissolving PVA in deionized water and heating to 90 °C followed by de-aeration to obtain a bubble free solution (Fig. 1a). The polyamide NF membrane of 600 MWCO

was dip coated in dilute PVA solution to obtain functionalized FNF membrane of 400 MWCO with a four layered structure possessing an overall thickness of 156 lm as shown in Fig. 1b.

196

B. Venkata Swamy et al. / Chemical Engineering Journal 233 (2013) 193–200

A similar procedure was followed to synthesize TFC polyamide membrane for RO except that the aqueous medium contained 1% meta-phenylenediamine (MPD) instead of piperazine. These indigenous membranes were scaled up into a spiral wound modules of 2.5 in. dia  21 in. long dimensions (Fig. 1c) with the help of Permionics Membranes Pvt. Ltd. 2.3. Description of pilot scale NF and RO systems The schematic diagram of pilot scale NF/RO system is shown in Fig. 2. A feed tank of 30 L capacity made of stainless steel was provided for storage and supply of the effluent to the system. A polypropylene (PP) prefilter cartridge of 5 lm pore size was installed upstream of the spiral wound membrane module to prevent the entry of suspended solid particles which could damage the membrane under pressurized condition. A high pressure pump (Hironisha, Japan) capable of maintaining a pressure up to 50 bar was installed for transporting the feed liquid thoughout the system. The pump was run by a 2 HP single phase motor (Crompton, India). The feed tank had a provision for recycle of the reject which passed though a heat exchanger (HE) for maintaining constant temperature (28–30 °C). Ice-cold water was circulated though the shell side of the HE whereas the effluent reject flowed though the tube side which consisted of concentric glass coils that provided large heat transfer area. The effluent coming out of the HE was then fed back to the feed tank as concentrate with total recycle. A restricting needle valve was provided on the concentrate outlet of the membrane pressure vessel at a position prior to the HE, to pressurize the feed to the desired value indicated by a pressure gauge installed upstream of the valve. Permeate and concentrate flow rates were measured using glass rotameters containing metal floats. 2.4. Experimental procedure Before starting the experiment, the NF/RO system was cleaned and wetted using distilled water until the permeate conductance reached 0.01 mS/cm. The experiments were carried out with deionized water to study the effect of pressure on pure water flux. Deionized water was taken in the feed tank and transported though the spiral wound NF/RO membrane module using high pressure pump. The concentrate tubing was left to flow outside into a bucket instead of feed tank in order to maintain constant feed concentration. The system pressure was varied by restricting the needle valve in the concentrate line. The flow rates of permeate and concentrate were recorded at each pressure.

Initially, the stainless steel feed tank was filled with 30 L of industrial effluent and the system was run to remove 2.2 L of distilled water that was present in the system, which is known as dead volume. A sample of initial feed was collected for analysis. The experiment was performed at 75% solvent recovery, optimum pressure of 21 bar (300 psi), with complete recycle of the concentrate to study the effect of feed concentration on flux and % rejection. The flow rates of permeate and concentrate were measured at regular time intervals to note any decline in flux. After a particular water recovery was attained, the initial feed, final concentrate and average permeate samples were analyzed for TDS, conductivity, COD, BOD and turbidity. Finally, the system was cleaned and washed with distilled water to remove solutes from the membrane surface and pores. 2.5. Fouling and its prevention Fouling of membranes is caused by suspended solids, microbes and organic materials present in the feed water that accumulate either on the membrane surface or within the pores [20]. An aqueous solution of citric acid or HCl (1% w/v) was run though the system for about 10–15 min for removal of mineral salt scales. Tetra sodium EDTA of 1% w/v and trisodium phosphate/sodium hydroxide was used to remove organic and inorganic scales. Finally, sodium lauryl sulfate was used for polishing the membranes by generating a soapy lather. 2.6. Analytical methods The feed and permeate samples were analyzed for TDS, COD, BOD and color (platinum–cobalt procedure) according to APHA methods [21]. The conductivity of above samples was determined using the digital conductivity meter. The feed and permeate samples were treated with mercuric sulfate (HgSO4), which converts the chlorides into insoluble mercuric chloride (HgCl2) precipitate and gets removed from the titration medium. 3. Results and discussion In NF and RO processes, the separation performance of the membrane is denoted in terms of flux and % rejection of TDS or any other feed constituent. The flux of solvent though the membrane can be described by [22]:

J w ¼ Aw ðDP  DPÞ

Fig. 2. Schematic representation of pilot-scale NF/RO system.

ð1Þ

B. Venkata Swamy et al. / Chemical Engineering Journal 233 (2013) 193–200

197

The % rejection is determined from:

  cp  100 %R ¼ 1  cf

ð2Þ

The approximate osmotic pressure P is determined from the following equation in order to determine the minimum pressure required to generate flux:

P ¼ ni C i RT

ð3Þ

3.1. Effect of feed pressure on flux The effect of feed pressure on flux of pure water and effluent for both NF and RO processes is shown in Fig. 3. As expected, a rise in feed pressure results in an increase of flux. With increase in feed pressure from 14 to 42 bar (200,600 psi), increment in flux was observed in the range 30–85.6 L/m2 h for NF and 18–48 L/m2 h in case of RO. Since the driving force of the process increases it results in enhancement of water flux due to increased affinity between H2O molecules and the polar –CONH groups of RO membrane or –OH functional groups in FNF. On the other hand, the sorption of solute molecules in the membrane remains more or less the same even at high pressures due to lack of any interaction. The flux was zero at applied pressures less than 7 bar (100 psi) for the effluent due to high osmotic pressure arising from substantial concentration of dissolved solids in the effluent feed. However, significant flux was obtained with pure water feed wherein trans-membrane pressure gradient (DP–DP) is quite high. 3.2. Effect of feed concentration on flux and % rejection Effect of feed TDS concentration on flux and rejection properties of both NF and RO membranes is graphically illustrated in Fig. 4. With increase in TDS content, a reduction in flux and% rejection was observed at a constant feed pressure of 21 bar (300 psi). Flux decreased from 47.2 to 28.3 L/m2 h during NF operation and from 39.2 to 8.1 L/m2 h in case of RO while the corresponding rejections reduced from 95% to 85% in NF and 98.5% to 90% in RO process, respectively, which is due to increasing concentration polarization near the membrane surface. Higher flux and lower rejection in case of NF may be attributed to the porous nature of the membrane that has pore size in the range 0.5–2 nm which is ten times higher than that in RO. 3.3. Separation of TDS, conductivity, turbidity, COD and BOD Table 2 depicts the values of TDS, conductivity, turbidity, COD and BOD present in initial feed, average permeate and final reject

Fig. 4. Effect of feed concentration on flux and rejection for NF and RO systems at 21 bar (300 psi).

for both RO and NF processes. RO process reduces the TDS, conductivity, turbidity, COD and BOD present in the feed effluent to a greater extent than NF. The permeate obtained during the process was analyzed as per the APHA standards which can be safely disposed into environment or recycled for utilization in agricultural activities or industrial cooling towers [21]. 3.4. Mathematical model A mathematical model for NF/RO system (Fig. 5) was formulated to calculate the following parameters: (i) Permeate flux, % rejection and permeate composition for given feed composition and operating conditions. (ii) Prediction of membrane system behavior under different operating conditions. 3.5. NF/RO process model The process model for NF/RO systems was developed by Mason and Lonsdale using statistical mechanical equations [23]. Basic transport equation for the species i of a multi-component solution is based on the fact that the resultant of driving force is permeate flux and can be written as: N X a0 Bo cj ui 1 ðui  uj Þ þ ¼  ðrT li  F i Þ  i ðrp  cFÞ RT cD D g DiM ij iM j¼1



N X cj T Dij rln T cD ij j¼1

ð4Þ

On the left-hand side of Eq. (4) two flux-related terms are given, one for solutesolute interactions and one for solutemembrane interactions. On the right hand side thee terms appear which pertain to various driving forces including isothermal diffusion which depends on concentration, pressure and forced diffusion, viscous flow and thermal diffusion. The above equation is similar to Stefan–Maxwell equation [24] for multicomponent diffusion. For binary mixtures the above equation gives two transport equations, one for each feed component as written below:

Fig. 3. Effect of pressure on flux of pure water and pharmaceutical effluent.

c2 u1 1 a0 B0 ðu1  u2 Þ þ ¼  Dl1  1 Dp RT cD12 D1M gD1M

ð5Þ

c2 u2 1 a0 B0 ðu2  u1 Þ þ ¼  Dl2  2 Dp RT cD21 D2M gD2M

ð6Þ

198

B. Venkata Swamy et al. / Chemical Engineering Journal 233 (2013) 193–200

Table 2 TDS, conductivity, turbidity, BOD and COD in NF and RO processes. Sample

TDS (ppm)

Conductivity (mS/cm)

Turbidity (FAU)

BOD (ppm)

COD (ppm)

RO permeate NF permeate RO reject NF reject

285 856 20,560 18,800

0.44 1.34 32.1 29.0

0 7 1090 1026

2.402 4.805 11.409 10.211

540 1080 15120 6300

where 1, 2 represent the components present in the solution. After further simplification we obtain:

and for concentrated solutions one obtains:

J 1 ¼ c1 u1 ¼ Lc11 Dl1  Lc12 Dl2  a1 c1 L0 Dp

ð7Þ

J 2 ¼ c2 u2 ¼ Lc21 Dl1  Lc22 Dl2  a2 c2 L0 Dp

ð8Þ

By changing the component fluxes (J1, J2) to volumetric flow J V and solute flux JS apart from elaborating the pressure gradients to osmotic pressure gradient Dpa and hydrostatic pressure gradient DP, we obtain the following form of equations for binary mixtures:

J V  V 1 J 1 þ V 2 J 2 ¼ Lp ðDp  rm Dpa Þ

ð9Þ

a

J s  ðC 1 V 1 ÞxDp þ c2 ½1  ðC 1 V 1 Þrs J V

ð10Þ

For dilute ideal solutions with constant transport coefficients, we have rv ¼ rs ¼ r, which thereby reduces the number of parameters from four to thee with the term (c1 V 1  1 and Dpa  RTðdc2 =dz Þ). The equation for solute flux then becomes:

J s ¼ C 01 ð1  rÞJ V þ

Pe ¼ P¼

ð19Þ

Therefore, J v ¼ ðD1 C w þ D2 ÞDP m : Due to concentration polarization phenomena caused by selective permeation of solvent, the solute concentration in the membrane interface on the high-pressure side Cw is higher than its concentration in the bulk solution Cb. The surface concentration cannot be measured but can be calculated using Eq. (23) reported by Rautenbach and Albrecht [25].

C w ¼ C b þ ðC b  C p ÞðeJv =k  1Þ

ð20Þ

where k is mass transfer coefficient calculated from the available correlation, sh = f (Re, Sc). According to this theory, r = R which means that Eq. (17) can be written as:

DP m ¼ DP  R DP

ð21Þ

where the osmotic pressure is given as:

ð1  rÞJ V ðc02  c002 Þ g Pe  1

ð11Þ

DP ¼ ðC w  C p ÞRg T

ð22Þ

and

ð1  rÞJ V P

ð12Þ

xRT

ð13Þ

Dz

Fractional solute rejection R for NF/RO operation is given by:

P ¼1

Lp ¼ ðD1 C w þ D2 Þ DZ

c002 c02

ð14Þ

which on further simplification becomes:

R¼

rðele  1Þ ePe  r

ð15Þ

Jm ¼

Lp DP m Dz

ð16Þ

DP m ¼ DP  r DP

DPm ¼ DP  RðC w  C p ÞRg T

and C p ¼ C w ð1  RÞ

ð23Þ

therefore,

DPm ¼ DP  R2 Rg TC w

ð24Þ

The developed statistical mechanical model was used to minimize the number of experiments to be performed and also to aid the design of a commercial NF/RO unit. The variation in the pressure, concentration and mass transfer coefficient of feed and permeate streams along the length of the membrane module can be calculated using this one-dimensional model. This model can also be extended to a two-dimensional model, wherein the changes in

ð17Þ Pe

Using Eqs. (12) and (15) and applying power series to e , we get:

1 c1 ¼ þ C2 R Jm

Fig. 5. Process Flow diagram of NF/RO commercial system.

ð18Þ

Fig. 6. Simulated results depicting effect of feed concentration on flux and rejection at 13.8 bar.

199

B. Venkata Swamy et al. / Chemical Engineering Journal 233 (2013) 193–200 Table 3 List and capital equipment cost for NF and RO systems. Item

Capacity/Size

MOC

Quantity

Membrane housing TFC RO membrane FNF-400 membrane 1354 Pressure vessel Skid Filter assembly Feed pump (2.5 HP) 1 HP high pressure pump 40 Nb TMF multiport valve Online TDS meter Sand and pebbles Cleaning pump 3-Phase control panel Carbon bags Hardware lot Ozonator UV system Total cost

1 m long (2.500 dia  4000 long) (2.500 dia  4000 long) 35 lpm 17 lpm 35 lpm 35 lpm 35 lpm – – – 2 lpm – – – – –

– PA PA – SS PP – SS – – – SS316 SS 350/500 – UPVC – –

1 1 1 1 1 2 1 1 2 1 3 Bags 1 1 2 1 set 1 1

Cost (USD) Unit

Total

190 300 300 200 200 7.5 116 1200 60 79 10 116 170 60 200 300 300

190 300 300 200 200 15 116 1200 120 79 30 116 170 120 200 300 300 $ 3656

PP – polypropylene, PA – polyamide, SS – stainless steel, UPVC– unplasticized polyvinyl chloride.

Table 4 Operation and maintenance cost for NF and RO systems. NF

RO

Feed capacity (m3/h) Permeate capacity (m3/h) Recovery offered Module replacement cost

2 1.8 90% 300

2 1 50% 300

Duration of replacement (Years) No. of working h/day Cost/h (USD)

5 12 0.0137

3 22 0.0124

Cartridge replacement cost No of cartridges (NF/RO) Total cartridge replacement cost (@ 8 USD) Duration of replacement (days) Cost/h (USD)

2 16 90 0.015

3 24 90 0.012

Power cost Feed pump (kW) Dosing systems (kW) High pressure pump (kW) UV lamp (kW) Ozonator (kW) Total power consumption-kW Hourly cost (@ 0.1 USD)

0.3725 0 0.3725 0.072 0.1 0.917 0.0825

1.12 0.015 1.85 0.072 0.1 3.157 0.284

Chemical consumption Antiscalant dosing (ppm) Dosage (L/h) Cost/L (USD) Hourly cost (USD)

5 0.01 6.426 0.064

5 0.01 6.426 0.064

15 0.0452 0.22 963.6

15 0.0452 0.4176 3353.3

335.6 2500 3799.2

365.6 2660 6378.9

1800 12 7884 0.48

1000 22 8030 0.7943

27736 0.1369

25741.1 0.2478

CIP chemicals (EDTA, NaOH, Citric acid) Frequency (days) Total cost of CIP per hour-(USD) Total operating cost per hour-(USD) Total operating cost per year assuming 22 h & 12 h of operation/day for RO & NF. (USD) Depreciation cost (Assuming 10% of capital cost) (USD) Labor cost per year + Raw water (USD) Total cost per year (USD) Permeate Quantity (L/h) Operation time (h) Quantity of permeate generated in 1 yr (kL/yr) Cost/kL of permeate (USD) If sold at 4  103 USD/L Annual profit (USD) Pay back period (yr)

solute concentrations on feed and permeate side along and across the length of flow could be predicted in order to improve design to

achieve better membrane performance. The model could also be useful in predicting the performance of other hydrostatic pressure-driven membrane pilot plants and commercial systems. 3.6. Model validation for RO/NF In order to evaluate the accuracy of the statistical mechanical model, it has been validated using the experimental data generated in the laboratory for bulk drug industrial effluent obtained from GVK Biopharma Ltd., Hyderabad. In this process, the reject was completely recycled to the feed tank. Therefore, the concentration of dissolved solid particles in the feed tank increased with time during both NF and RO trials. Fig. 4 represents the comparison of theoretically predicted results with experimental values especially with respect to reject stream composition, % solute rejection and flux. The experimental results obtained were in good agreement with theoretical values. Model presented in Appendix A was developed in MATLAB and simulated using NF experimental data. The regression coefficients were found to be C1 = 4.5955e6, C2 = 0.7142, D1 = 2.2641e018 and D2 = 8.6010e16. The maximum average error was found to be 3% for flux and 1.14% for % rejection. Similarly, RO phenomenon was simulated and the regression coefficients were determined to be C1 = 1.6829e7, C2 = 1.0021, D1 = 2.1789e17 and D2 = 7.8141e16. The flux and rejection were calculated using the model coefficients and the maximum average error was found to be 4% for flux and 0.7% for rejection. 3.7. Simulation During simulation study, the effect of different operating parameters such as feed concentration and feed pressure on membrane system performance was investigated and outputs are shown in Fig. 6. The simulation results thus obtained exhibited concurrence with experimental data. 3.8. Scale-up and economic estimation 3.8.1. Equipment list and capital investment for NF and RO systems List of equipments and capital costs are provided in Table 3, in which the unit price for the high-pressure centrifugal pump (Grundfos, Denmark), FNF-400 and TFC Polyamide RO membrane modules with pressure vessel and skid costs are included.

200

B. Venkata Swamy et al. / Chemical Engineering Journal 233 (2013) 193–200

3.8.2. Operation and maintenance cost in NF and RO processes Operating and maintenance cost of NF and RO systems are given in Table 4, which includes membrane and prefilter cartridge replacement costs, electric power consumption besides chemicals for cleaning and storage of membranes. Feed capacity and recovery were assumed to be 2 m3/h and 50% for RO, whereas in case of NF, the corresponding values were taken as 2 m3/h and 90%. The operating durations for RO and NF systems were assumed to be 22 and 12 h per day correspondingly. Depreciation costs were taken as 10% of the capital investment.

After calculation of rejection and flux, feed concentration can be calculated using material balance equations as follows: Feed input to the RO module = Permeate collected + Reject Recycle i.e. F ¼ R þ P

Feed balance : QðtÞ ¼ Q ð0Þ  J  area  time

ð25Þ

Solute balance : Q ðtÞCðtÞ ¼ Qð0ÞCð0Þ  J  area  time  C p

ð26Þ

Using Eqs. (25) and (26) we can calculate concentration as a function of time as well as total duration of operation.

4. Conclusions

References

The study reveals that functionalized nanofiltration and thin film composite reverse osmosis membranes can be effectively used for the treatment of bulk drug industrial effluent. The flux, rejection and permeate composition for a given feed was evaluated under various operating conditions. With the obtained results a mathematical model based on statistical–mechanical transport equations was adapted to design commercial membrane plants. The simulation results obtained showed a very good agreement with only 5% deviation from experimental data. Scale-up aspects and economics of nanofiltration and reverse osmosis systems were assessed. From experimental results it can be concluded that nanofiltration is more economical than reverse osmosis apart from sharing the same advantages such as ecofriendly nature, process safety, ease of operation and maintenance besides small footprint albeit with slight limitations in separation efficiency.

[1] A.R.D. Verliefde, E.R. Cornelissen, S.G.J. Heijman, I. Petrinic, T. Luxbacher, G.L. Amy, V.B. Bruggen, J.C. Van Dijk, Influence of membrane fouling by (pretreated) surface water on rejection pharmaceutically active compounds (PhACs) by nanofiltration membranes, Journal of Membrane Science 330 (2009) 90–103. [2] Y. Yeomin, W. Paul, A.S. Shane, C.W. Eric, Nanofiltration and ultrafiltration of endocrine disrupting compounds, pharmaceuticals and personal care products, Journal of Membrane Science 270 (2006) 88–100. [3] A.A. Abdulhakeem, T. Nichanan, J.A. Mc-Donald, J.K. Stuart, E.P. William, D.N. Long, Combining MBR and NF/RO membrane filtration for the removal of trace organics in indirect potable water reuse applications, Journal of Membrane Science 365 (2010) 206–215. [4] S. Halling, S.N. Nielson, P.F. Lanzky, F. Ingerslev, J.H. Lutzhoft, S.E. Jorgensen, Occurrence, fate and effects of pharmaceutical substances in the environment – a review, Chemosphere 35 (1998) 357–393. [5] K. Ismail, A.A. Osman, R.W. Mark, C. Rice, Removal of hormones and antibiotics by nanofiltration membranes, Journal of Membrane Science 309 (2008) 94– 101. [6] S. Khetan, T. Collins, Human pharmaceuticals in the aquatic environment: a challenge to green chemistry, Chemical Reviews 107 (2007) 2319–2364. [7] H.C. Flemming, G. Schaule, T. Griebe, J. Schmitt, A. Tamachkiarowa, Biofouling – the achilles heel of membrane processes, Desalination 113 (1997) 215–225. [8] Q. Elimelech, M. Li, Synergistic effects in combined fouling of a loose nanofiltration membrane by colloidal materials and natural organic matter, Journal of Membrane Science 278 (2006) 72–82. [9] D. Squire, Reverse osmosis concentrate disposal in the UK, Desalination 132 (2000) 47–54. [10] D.N. Long, I.S. Andrea, Critical risk points of nanofiltration and reverse osmosis processes in water recycling applications, Desalination 187 (2006) 303–312. [11] M. Mickley, Directions in management of membrane side streams (Part II), Membrane Technology 2003 (8) (2003) 8–12. [12] G.A. Schader, A. Zwijnenburg, M. Wessling, The effect of WWTP effluent zeta– potential on direct nanofiltration performance, Journal of Membrane Science 266 (2005) 80–93. [13] Y.Q. Victor, K.M. Sung, F. Takahiro, K. Maria, A. Gary, Proposing nanofiltration as acceptable barrier for organic contaminants in water reuse, Journal of Membrane Science 362 (2010) 334–345. [14] Z. Amjad (Ed.), Reverse Osmosis: Membrane technology, Water chemistry, and Industrial Application, Van Nostrand Reinhold, New York, 1993. [15] S.K. Jain, M.K. Purkait, S. De, P.K. Bhattacharya, Treatment of leather plant effluent by membrane separation processes, Separation Science and Technology 41 (2006) 3329–3348. [16] T. Tinghui, T. Matsuura, S. Sourirajan, Effect of membrane materials and average pore size on reverse osmosis separation of dyes, Industrial & Engineering Chemistry Product Research Development 22 (1983) 77–85. [17] A. Erswell, C.J. Brouchaert, C.A. Buckley, The reuse of reactive dye liquors using charged ultrafiltration membrane technology, Desalination 70 (1988) 157– 167. [18] S. Sridhar, A.G. Rao, Treating effluents in bulk drug industry using reverse osmosis, Pharma Bioworld, 2004. [19] S. Sridhar, B. Smitha, Performance and prospects of nanofiltration for separations in pharmaceutical manufacture, Management/Environment, Pharma Bioworld, 2004. pp. 78–83. [20] J. Kucera, Properly apply reverse osmosis, Chemical Engineering Progress 93 (1997) 54–61. [21] Standard methods for examination of water and wastewater, fourteenth ed., APHA-AWWA-WPCF, USA, 1976. [22] G. Owen, M. Bandi, J.A. Howell, S.J. Churchouse, Economic assessment of membrane processes for water and wastewater treatment, Journal of Membrane Science 102 (1995) 77–91. [23] E.A. Mason, H.K. Lonsdale, Statistical-mechanical theory of membrane transport–a review, Journal of Membrane Science 51 (1990) 1–81. [24] R. Rautenbach, R. Albert, Membrane Processes, John Wiley and Sons, New York, 1989. pp. 29–30. [25] C. Muckenfuss, Stefan–Maxwell relations for multicomponent diffusion and the Chapman Enskog solution of the Boltzmann equations, Journal of Chemical Physics 59 (1973) 1747–1752.

Acknowledgments We are thankful to Council of Scientific and Industrial Research (CSIR, New Delhi) for granting funds though MATES XII Five Year Plan project to support our research activities and BVRIT based at Medak District for providing MATLAB 7.0 facility.

Appendix A. Calculation procedure assuming existence of concentration polarization (U)