Energy-efficient reverse osmosis desalination process

Highlights

• New desalination process combines SSRO with countercurrent membrane cascade.

• Multistage process requires no interstage pumping on retentate side.

• Reduced osmotic pressure difference relative to conventional SSRO at all recoveries.

• Critical recovery above which SEC is less than that for conventional SSRO.

• 75% recovery with 33% reduction in pressure and 11% in SEC relative to SSRO.

Abstract

A novel energy-efficient reverse osmosis (EERO) process is proposed for which the retentate from single-stage reverse osmosis (SSRO) serves as the feed to a countercurrent membrane cascade with recycle (CMCR). The 3-stage EERO process employs two, whereas the 4-stage EERO process employs three stages in the CMCR. The EERO process is advantageous because of four features: (i) coupling SSRO with a CMCR; (ii) countercurrent retentate and permeate flow; (iii) permeate recycling; and (iv) retentate self-recycling owing to the use of one or more nanofiltration stages. The EERO process was compared to conventional SSRO for both processes operating at the thermodynamic limit and employing an energy-recovery device. For the same overall recovery the osmotic pressure differential is reduced by 33% and 50% relative to SSRO for the 3- and 4-stage ERRO processes, respectively. There is a critical recovery above which the EERO process also can reduce the specific energy consumption (SEC) relative to SSRO for the same recovery. For a typical seawater feed of 35 g/L the 3-stage EERO process can achieve a 75% recovery at a net SEC of 2.746 kWh/m³, an 11.0% reduction in the SEC relative to SSRO for the same recovery. The 4-stage EERO process can achieve a 75% recovery at the same net SEC as SSRO (3.086 kWh/m³). Accounting for the additional membrane area required for the EERO process increases its cost relative to that for SSRO by at most 8%. An additional benefit of the EERO process relative to SSRO is the highly concentrated retentate that reduces the brine disposal volume or can be used to greatly increase the draw potential to harvest its osmotic potential energy via the pressure-retarded osmosis process.

Introduction

There has been a continued thrust in membrane desalination to reduce the specific energy consumption, increase the water recovery, and reduce the osmotic pressure differential (the osmotic pressure difference between the retentate and permeate in a stage). In view of recent progress in developing high flux reverse osmosis membranes, more effort is now being focused on optimizing the process design for desalination [1], [2], [3]. A useful approach to improve membrane desalination process design is to draw from optimization strategies developed for separations processes such as distillation that employ multistage processing with recycling of one or both countercurrently flowing streams. Multistage distillation also involves bi-directional transfer in each stage whereby the less volatile components in the vapor condense and the more volatile components in the liquid vaporize, which improves the separation and enhances the recovery. Since permeation in membranes is unidirectional, the advantages of bidirectional transfer in each stage must be achieved by interstage recycling of permeate to the retentate side and either by direct transfer of retentate to the permeate side or by employing membranes with only moderate rejections, which will be referred to as ‘retentate self-recycling’.

Some specialized terminology necessarily is used quite often in this paper. Hence, abbreviations have been introduced to make the paper more readable, a list of which is provided in the paper.

A countercurrent membrane cascade with recycle (CMCR) will cause the salt concentration to increase in the direction of the retentate flow. This minimizes the concentration difference across the membrane and thereby reduces the osmotic pressure differential (OPD) in each stage. The OPD can be further reduced by recycling retentate to the permeate side and permeate to the retentate side. Reducing the OPD translates to a direct reduction in the pumping costs as well as reduced maintenance costs. Lower pressure operation will also reduce the fixed costs vis-à-vis the expense for constructing a desalination facility. Introducing the feed optimally into a stage in the CMCR that does not directly produce the potable water product divides the CMCR into purifying and stripping sections that will increase the purity and recovery of the water product. It also can reduce the fouling in the reverse osmosis (RO) stage that produces the water product. However, these advantages of a CMCR are accompanied by shortcomings. Recycling increases the pumping requirements and may require interstage booster pumps. Moreover, the increased stream flows will require a larger membrane area. Multistage processing necessarily involves an increase in the complexity that complicates the process control.

The focus of this paper is to explore the advantages of a novel energy-efficient reverse osmosis (EERO) process that combines single-stage reverse osmosis (SSRO) with either a 2-stage or 3-stage CMCR. The metrics for this assessment are the overall water recovery, OPD, membrane area, and specific energy consumption (SEC) without and with an energy-recovery device (ERD).

The EERO process involves an optimal combination of the SSRO stage with a CMCR that employs countercurrent retentate and permeate flow, permeate recycle and retentate self-recycling. These elements of the EERO process and the reason for employing them are reviewed here.

It is appropriate to begin with a review of SSRO since it is a component of the EERO process and is the basis of comparison for assessing improvements in the recovery, OPD, and SEC. Fig. 1 is a schematic of SSRO in which the dotted line shows a retentate recycle stream; $Q_i\mathrm{and}{C}_i,$ denote the flow rate and concentration of stream i. Consider first SSRO in the absence of retentate recycle or any ERD, without any concentration polarization, and operating at the thermodynamic limit (at a transmembrane pressure equal to the OPD between the retentate and permeate in a stage). For specified feed and permeate concentrations the OPD and SEC are unique functions of the water recovery, $Y_{\mathrm{SSRO}}\equiv Q_0/Q_{\mathrm{f}},$ given by

$$Y_{\mathrm{SSRO}}=1-\frac{K}{\mathrm{\Delta }\pi }\left(C_{\mathrm{f}}-C_0\right)$$

$$\mathrm{SEC}=\frac{K\left(C_{\mathrm{f}}-C_0\right)}{Y_{\mathrm{SSRO}}\left(1-Y_{\mathrm{SSRO}}\right)}$$

where the subscripts f and 0 denote the feed and permeate, respectively and K is the coefficient in the linear relationship between the concentration and osmotic pressure. Eq. (2) implies a minimum SEC when $Y_{\mathrm{SSRO}}=0.5$ at which the OPD and SEC are given by

$$\mathrm{\Delta }\pi =2K\left(C_{\mathrm{f}}-C_0\right)$$

$$\mathrm{SEC}=4K\left(C_{\mathrm{f}}-C_0\right)$$

The OPD and SEC given by Eqs. (3), (4), respectively, will be the benchmarks for assessing the performance of the EERO process. For example, for a typical 35 g/L seawater feed and 0.35 g/L water product, the minimum $\mathrm{SEC}=3.086\mathrm{kWh}/{\mathrm{m}}^3$ at $\mathrm{\Delta }\pi =55.5\mathrm{bar}$.

The water recovery of conventional SSRO can be increased by recycling some retentate back to the feed as shown by the dotted line in Fig. 1. The corresponding OPD and SEC (without an ERD) are given by

$$\mathrm{\Delta }\pi =\frac{K\left(C_{\mathrm{f}}-C_0\right)}{1-Y}=\frac{K\left(C_{\mathrm{f}}-C_0\right)\left[\chi \left(1-R\right)+1\right]}{\chi \left(1-R\right)}$$

$$\mathrm{SEC}=\frac{K\left(C_{\mathrm{f}}-C_0\right)}{Y\left(1-Y\right)}=\frac{K\left(C_{\mathrm{f}}-C_0\right){\left[\chi \left(1-R\right)+1\right]}^2}{\chi \left(1-R\right)}$$

where $R\equiv Q_1/Q_3$ is the retentate recycle ratio and $\chi \equiv Q_3/Q_0$ is the safety factor (the ratio of the retentate to permeate flow). Desalination processes are typically operated with χ ≥ 1 to minimize membrane fouling. Eqs. (5), (6) for the OPD and SEC in terms of Y are the same as Eqs. (1), (2); hence, the OPD, SEC and recovery are unchanged from those for conventional SSRO when SSRO with retentate recycle is operated optimally at the minimum SEC. Eqs. (5), (6) indicate that the minimum SEC can be achieved for various combinations of R and χ. For a specified R there is a unique χ that will give the same minimum SEC as SSRO given by

$$\chi =\sqrt{\frac{1}{1-2R+R^2}}$$

Eq. (7) indicates that χ=1 at R=0 and $\chi \to \infty$ as $R\to 1$. Operating at higher χ is advantageous for mitigating fouling since increased flow on the retentate side reduces concentration polarization. However, retentate recycle increases the recovery at the expense of an increased SEC relative to conventional SSRO.

The recovery also can be increased by using two RO stages in series whereby the retentate from the first stage becomes the feed to the second stage and the permeate streams are combined to give the product water as shown in Fig. 2. Since the retentate has a higher concentration than the feed in stage 1, a booster pump is needed for additional recovery from stage 2. The OPD for stage 2, overall recovery, and gross SEC (without an ERD) for the 2-stage configuration when stage 1 is operated optimally (i.e., at $Y_1=0.5$) are given by

$$\mathrm{\Delta }{\pi }_2=\frac{\mathrm{\Delta }{\pi }_1}{1-Y_2}$$

$$Y=\frac{1}{2}\left(1+Y_2\right)$$

$$\mathrm{SEC}=\frac{\mathrm{\Delta }{\pi }_1\left(1-(1/2)Y_2\right)}{1/2\left(1+Y_2\right)\left(1-Y_2\right)}$$

where $\mathrm{\Delta }{\pi }_1$ and $\mathrm{\Delta }{\pi }_2$ are the OPD in stages 1 and 2, respectively, $Y_2$ is the recovery in stage 2 and Y is the overall recovery. The SEC for this 2-stage series configuration is a minimum when $Y_2=0.268$ at which the overall recovery is 63.4% and the gross SEC is 2.88 kWh/m³; however, this comes at the cost of an OPD in the second stage of 75.8 bar. For an overall recovery of 75% the 2-stage series configuration has a gross SEC of 3.09 kWh/m³ and requires an unreasonably high OPD in the second stage of 111 bar. Hence, the 2-stage series configuration is not viable for achieving high water recoveries.

Another way to increase the recovery is to employ a CMCR. Fig. 3 shows a generalized CMCR process with n total stages, countercurrent retentate and permeate flow, permeate recycle in each stage, and introduction of the feed optimally between stages. The latter divides the CMCR into retentate-enriching and permeate-enriching sections containing m and n–m stages, respectively.

The CMCR dates back to the 40s when it was used for the gas-phase separation of U²³⁵ and U²³⁸ for which the membranes had a very low rejection [4]. A CMCR is frequently used for gas separations for which highly rejecting membranes are unavailable. Pathare and Agrawal [5] considered a CMCR for producing nitrogen from air and found that there was an optimum number of stages for a specified retentate product purity and recovery that minimizes the SEC. This arises because permeate recycling reduces the OPD but also increases the gas flow that must be repressurized.

CMCRs have also been used for separating the components in liquid feeds for low or moderately rejecting membranes. Siew et al. [6] analyzed a CMCR for separating an active pharmaceutical ingredient from an organic solvent. They used moderately rejecting nanofiltration (NF) membranes; hence, retentate self-recycling occurred in each stage. Having purifying and stripping sections with both interstage permeate recycling and retentate self-recycling permitted good recovery of both components. Adding stages increased the fixed costs, whereas permeate recycle increased the operating costs.

The reflux reverse osmosis (REFRO) process of Loeb and Block [7] was the first use of a CMCR for desalination. This employed a 2-stage CMCR with the feed introduced at one end, permeate recycle from the second to the first stage, and retentate recycle from the high to the low pressure side of the second stage. REFRO increased the recovery and reduced the OPD owing to the permeate and retentate recycling. However, the SEC for REFRO is higher than that for conventional SSRO.

Ting et al. [8] considered a 3-stage CMCR that was first proposed by Dobbs [9]. They employed loose NF, tight NF, and RO membranes in stages 1, 2, and 3, respectively, with interstage permeate recycling and the feed introduced between stages 2 and 3 (see Fig. 3). They considered a 32 g/L saline water feed and specified the membrane rejections. The NF membranes permitted self-recycling in stages 1 and 2. They did not determine the minimum SEC but only reported gross SEC values of 3.32, 4.49 and 4.75 kWh/m³ at recoveries of 10%, 15% and 20%, respectively. These SEC values are below those for SSRO operating at these recoveries but above those for optimal operation of SSRO. When operated optimally for the feed and product concentrations of Ting et al., SSRO has a 50% recovery at an SEC of 2.80 kWh/m³. Moreover, the OPD values reported by Ting et al. range from 1.8 to 2.5 bar in the RO stage, 9.9 to 13 bar in the tight NF stage, and 23 to 26 bar in the loose NF stage, thereby requiring two high pressure booster pumps. Hence, the 3-stage CMCR process of Ting et al. does not compare favorably with conventional SSRO.

Karuppiah et al. [10] reviewed optimization studies for determining the membrane process configuration(s) that minimize the total cost for desalination. Most of these studies are variations of the optimization methodology of El-Halwagi [11] and include those of Voros and Maroulis [12], [13], Voros et al. [14], Zhu et al. [15], Van der Meer [16], [17], Wessels and Van der Meer [18], Lu et al. [19], [20], Vince et al. [21], and others. These optimization schemes require specifying the membrane properties in order to determine the total costs. As such, they do not permit exploring the optimum rejections for each stage. The focus of these multistage optimization studies is also different in that they are not concerned with minimizing the OPD or increasing the recovery.

Conventional SSRO, when operated optimally, has a reasonable recovery for a typical seawater feed but at a high OPD. Retentate recycle can increase its recovery at the expense of an increased SEC. Employing RO stages in series can increase the recovery and decrease the SEC but requires interstage booster pumps. A CMCR employing moderately rejecting membranes in some stages can increase the recovery and reduce the OPD. However, the increased pumping costs for repressurizing the permeate recycle increases the SEC. This overview suggests that an optimal desalination process would combine the benefits of conventional SSRO with those of a CMCR. Hence, the focus of this study was to develop a novel membrane process configuration that could increase the recovery, reduce the OPD and SEC relative to the prior art.