Energy consumption for desalination — A comparison of forward osmosis with reverse osmosis, and the potential for perfect membranes

i n f o

A customised simulation tool was used to estimate the SEC for RO and FO desalination by considering the effect of different process variables and UF pretreatment step. Using a CAPE-OPEN interface standard for running Matlab scripts in ASPEN, this modelling approach provided a flexible tool for quantifying the energy consumption of desalination by simulating real process conditions. It was concluded that there is effectively no difference in SEC between the FO with NF recovery and RO processes. Furthermore, it has been shown that even if any of the membranes, FO, RO or NF had infinite permeabilities and 100% rejection, it would not change the SEC significantly. Based on these simulations alone, FO with NF recovery cannot be considered to be competitive with RO taking into account the additional capital costs needed for FO with NF recovery, unless other advantages of the process can be capitalised on. One such advantage is the apparently lower fouling propensity of FO [3,59–64] which may reduce or eliminate the need for pretreatment and chemical cleaning, thus reducing costs. In order to investigate if this phenomenon can be exploited, the mechanism and extent of fouling in FO compared to RO needs to be further studied and understood. The FO–Distillation process with CO2–NH3 draw solution showed the lowest SEC compared to other FO and RO desalination processes. However, concerns over residual NH3 being above the allowable limit in the product water is a challenge which remains to be resolved. At 75% recovery, the single-stage FO–UF process with NPs as the draw solution is estimated to have a similar SEC to the FO with singlestage NF process. A two-stage UF for the nanoparticle recovery may result in similar SEC as the two-stage RO and FO with two-stage NF recovery processes, albeit increasing capital costs. The lack of data in the literature makes it challenging to model the SEC for this process at varying product recoveries and process conditions. Hence, more research is required in this area to increase the availability of data before accurate comparisons can be made with other desalination processes. In general, it was observed that despite the type of draw solution and pressure-driven recovery method used, there is effectively no difference in energy consumption of different hybrid FO processes and the standalone RO process. This is because, the requirement for πDS Recovery, Brine = πFO ,Draw negates the benefit of using draw solutes which can be recovered by low pressure processes. This analysis can be generalised for any pressure-driven membrane process used for the DS recovery stage, although there are still opportunities for hybrid FO processes to provide energy cost savings by leveraging on low-cost thermal energy DS recovery methods such as the FO–Distillation process for recovering the CO2–NH3 DS. Nomenclature carbon dioxide CO2 CP concentration polarisation DS draw solution ERD energy recovery device ESI electronic supplementary information FO forward osmosis FS feed solution HPP high pressure pump HTI Hydration Technology Innovations LPP low pressure pump MBR membrane bioreactor MNP magnetic nanoparticles ammonia NH3 NP nanoparticles PAA polyacrylic acid PAA-NP polyacrylic acid-nanoparticles RO reverse osmosis SEC specific energy consumption SW module spiral wound module SWRO seawater reverse osmosis UF ultrafiltration UPP ultrafiltration permeate pump WHO World Health Organisation P hydraulic pressure (Pa) π osmotic pressure, Pa ΔP transmembrane pressure difference, Pa Δπ osmotic pressure difference across the membrane, Pa osmotic pressure difference between the permeate and feed Δπ1 side for the FO stage, Pa osmotic pressure difference between the draw and retentate Δπ2 side for the FO stage, Pa corrected van't Hoff factor for NaCl solution = 1.64 ∅NaCl ∅MgSO4 corrected van't Hoff factor for MgSO4 solution = 1.2 feed flowrate, mol (solvent + solute)·s−1 QF QP permeate flowrate, mol (solvent + solute)·s−1 QPP pure water permeate flowrate in FO, mol (solvent + solute)·s−1 QR retentate flowrate, mol (solvent + solute)·s−1 QD draw solution flowrate, mol (solvent + solute)·s−1 Lp membrane permeance, m3·m−2⋅Pa−1⋅s−1 Am membrane area, m2 Am,s specific membrane area, [m2(m3·h−1)−1] T temperature, K R membrane rejection, % universal gas constant = 8.314 J·K−1·mol−1 Rg k external mass transfer coefficient, m·s−1 K solute resistivity for diffusion within porous support layer, s·m−1 Xi molar fraction of the solute νi (P, T, Xi) molar volume of mixtures and solvent as a function of pressure, temperature and molar fractions of the solute, supplied by the thermophysical properties engine, m3·mol−1 Pf feed side pressure, Pa ρ fluid density, kg·m−3 νf fluid velocity, m·s−1 dh hydraulic diameter of the feed channel, m λ friction factor factor to take into consideration pressure losses in the feed Kλ tubes and module fittings Wpump pump work, W equivalent work, kWh·m−3 Weq Re Reynolds Number x position down the feed flow path Y product recovery GSEC fES PSMC mnorm m mA β ε gain in energy savings fractional energy savings penalty due to the increase in membrane area dimensionless membrane price amortised membrane price per unit area, Pa·m3·m−2·h−1 amortised membrane unit cost, £·m−2·h−1 conversion factor, Pa·m3·kWh−1 energy price, £·kWh−1