Elsevier

Journal of Cleaner Production

Volume 140, Part 3, 1 January 2017, Pages 1429-1443
Journal of Cleaner Production

Wastewater treatment by spiral wound reverse osmosis: Development and validation of a two dimensional process model

https://doi.org/10.1016/j.jclepro.2016.10.008Get rights and content

Highlights

  • A two dimensional model applicable for spiral-wound RO system is developed.

  • Two-dimensional model can precisely estimate the performance of RO process.

  • The model showed agreement with the data of binary mixture wastewater experiments.

  • Spiral-wound RO module of 7.9 m2 can achieve desired dimethylphenol rejection.

Abstract

Reverse osmosis (RO) has become a significant method for removing salts and organic compounds from seawater and wastewater in recent decades. Spiral-wound module has been widely used due to a number of special features such as high packing density, premium separation and low operating cost. In this paper, a two-dimensional mathematical model is developed for the transport of dilute aqueous solutions through a spiral-wound RO module and the operational characteristics of the process under steady state conditions are analysed. The model is based on the solution-diffusion model coupled with the concentration polarization mechanism. This model yields a set of Differential and Algebraic Equations (DAEs), which are solved using the gPROMS software. The model is validated using experimental data from the literature for the rejection of dimethylphenol as solute in aqueous solutions. The model is then used to simulate the process under steady state conditions to gain deeper insight of the process.

Introduction

Reverse osmosis (RO) is a pressure driven process used to remove salts/pollutants and purify water so, it can be used for various purposes such as human consumption, agricultural and industrial use (Sassi and Mujtaba, 2011). Spiral-wound membrane modules are often preferred in both desalination and industrial processes since it offers specific characteristics of accepted permeation rates, low energy requirements, low fouling levels, ease of operation and low water production costs (Evangelista, 1988). Generally, RO is considered as a less expensive technology in comparison with ultrafiltration (Chew et al., 2016). This technology also showed a dramatic growth for water recycling and wastewater treatment in several industries (Lee and Lueptow, 2001). For example, RO is widely considered for wastewater/effluent treatments in (a) textile industry (Amar et al., 2009) (b) dairy industry (Koyuncu et al., 2000, Álrez et al., 2002) (c) tannery industry (Bhattacharya et al., 2013) and (d) pharmaceutical industry (Mitra-Gholami et al., 2012). These have stimulated continued research with an ultimate objective of maximizing the performance of the unit and reducing the cost of filtration with alleviation of environmental impact. For this purpose, a detailed but accurate process model is highly desirable, so that a reasonable prediction of the membrane performance can be obtained with feasible operating conditions for facilitating the best design of RO process.

There are a number of one and two-dimensional models in the literature developed to research the membrane performance of a spiral-wound module with different features and applications based on some assumptions and validated with sea and brackish water experimental data. Having said this, most of the suggested models have assumed constant pressure in a permeate channel (Karabelas et al., 2014). A critique on current literature is discussed in the following section.

A lumped model was developed by Gupta (1985) for a spiral-wound module under laminar and turbulent flows and based on the solution-diffusion model. It presumes constant mass transfer coefficient and solute concentration at the feed channel and neglects solute concentration at the permeate channel.

Analytical models were developed by Rautenbach and Dahm (1987) for a spiral-wound module and worked out by Evangelista (1988), for high rejecting membranes and by Avlonits et al. (1991) and Boudinar et al. (1992) for both Roga and FilmTech membrane types respectively. These models considered the validity of the solution-diffusion model with a fully axial flow of brine solution and neglected the components of the tangential feed flow and the axial permeate flow. Also, these models assumed constant density and viscosity and ignored the concentration polarization impact. In addition to this, some of these models did not consider the pressure drop in the brine and permeate compartments.

Based on the three parameter model of the Spiegler and Kedem, 1966, Senthilmurugan et al., 2005 and Mane et al. (2009) have developed models for turbulent flow by considering the pressure drop in both the channels. In comparison, Mane et al. (2009) have considered two dimensions (x and y) for the feed flow rate and stimulated the rejection of boron by the RO process.

Geraldes et al. (2005) have developed a one-dimensional model for spiral-wound RO membranes by ignoring both pressure drop in the permeate channel and the diffusion flow in the feed channel. While, Sagne et al. (2009) have considered a modified unsteady state one-dimensional model albeit by neglecting the concentration polarization impact and degrading the solute flux.

Avlonits et al. (2007) have developed a two-dimensional model by assuming only convective flux and neglecting diffusive mass transport and ignoring the variance of permeate concentration along the axial and spiral directions.

Oh et al. (2009) have developed a one-dimensional model based on the solution-diffusion model for spiral-wound RO system. It assumes constant mass transfer coefficient and constant water flux in the case of changing the inlet feed flow rate. Also, it neglects the pressure drop in the permeate channel. Kaghazchi et al. (2010) have proposed a one-dimensional model based on the solution-diffusion model where the bulk flow rate is calculated as an average value of inlet and outlet feed flow rates.

All the above models are validated with sea water and brackish water experimental data. In contrast, some comments on the proposed models that can be validated against wastewater treatment data are summarised in the next section.

A lumped model has been developed by Ahmad et al. (2007) for unsteady state simulation and validated with the experimental data of pre-treated palm oil mill effluent as a feed using a pilot plant scale RO system.

Sundaramoorthy et al., 2011a, Sundaramoorthy et al., 2011b have suggested a one-dimensional model by assuming the validity of the solution-diffusion model and constant values for both the permeate concentration and pressure along the permeate side. The model has been validated with the experimental data of chlorophenol and dimethylphenol solutes.

Fujioka et al. (2014) have also developed a one-dimensional model based on the irreversible thermodynamic model and considered the variety of the operating parameters by assuming zero permeate pressure. The model is validated against experimental data of N-nitrosamine rejection.

As can be seen from the above discussions that the models are restricted to one-dimension of spiral-wound RO process used especially for wastewater treatment and clearly neglect the tangential direction impact. Furthermore, Sundaramoorthy et al. (2011a) confirmed that there are only a few validation studies of mathematical models with wastewater experimental data. With the above backdrop, the initiative nature of this work lies in (a) the development of an explicit simple two-dimensional spiral-wound RO model applicable for dilute aqueous solution in wastewater treatment process (b) the validation of the model against experimental data from the literature and (c) the further application of the model to study the effect of various operating parameters on the performance of RO system. This model will be based on the solution-diffusion model and will relax the assumptions of constant physical properties, constant pressure and concentration of the fresh water on the permeate side considered in the past by many researchers. In addition, the brine concentration varies along the membrane length and width due to the impact of the plug-flow and diffusion flow. Also, it will consider the concentration polarization impacts on the whole unit.

Section snippets

The configuration of spiral-wound module

The configuration of the spiral-wound module essentially comprises a sealed envelope of membrane containing product water side and a spacer for the flowing of the feed. The membrane envelope is made of two sheets and sealed on three edges with an opening fourth edge connected with a central perforated pipe where the permeated water is collected. The narrow channels between the envelopes where the feed and permeate flow are filled with very thin fibers (spacers), which are wrapped around the

Model structure

To predict the characteristics of spiral-wound RO operation, the steady state mathematical model is suggested. From the above assumptions, the following sets of equations are formulated at any point in the system.

According to the solution-diffusion model, the characteristics of RO separation can be measured by the difference between the solvent and solute permeation fluxes Jw and Js through the membrane.

The solvent flux Jw(x,y) is proportional to the divergence between the hydraulic pressure

Model parameter estimation

The parameters of the model will be estimated by using the proposed graphical method of linear fit of Sundaramoorthy et al. (2011a). This method will be used to determine the values of solvent transport coefficient Aw, solute transport coefficient Bs and the feed channel friction parameter b. The details of these parameters are mentioned in Table 1. The estimated values of solvent transport parameters and solute transport parameters showed some difference than the values suggested by Srinivasan

Conclusions

A two-dimensional mathematical model applicable for dilute aqueous solution in a spiral-wound RO system has been developed and validated with a wastewater simulation study. The model can be used to predict the flow rate, concentration, pressure and temperature in each point along the two sides of the membrane length and width. Furthermore, this model facilitates the estimation of the behavior of water flux, solute flux and solute concentration on the wall of the membrane. A number of explicit

References (31)

Cited by (23)

  • Modeling and optimization of chlorophenol rejection for spiral wound reverse osmosis membrane modules

    2021, Chemosphere
    Citation Excerpt :

    On the other hand, various mathematical and thermodynamic approaches have been developed to optimize the spiral-wound RO process performance. For instance, A one-dimensional steady-state and dynamic mathematical model and a two-dimensional mathematical model have been used to dilute aqueous solution in the spiral-wound RO process (Al-Obaidi and Mujtaba, 2016; Al-Obaidi et al., 2017b). The limitations of mathematical models can overcome by computational methods developed to optimize the RO process much more quickly and accurately.

View all citing articles on Scopus
View full text