- Email: [email protected]
A method to find an optimal draw solute for cost-effective FO(forward osmosis) desalination process
I.A. Karimi and Rajagopalan Srinivasan (Editors), Proceedings of the 11th International Symposium on Process Systems Engineering, 15-19 July 2012, Singapore. © 2012 Elsevier B.V. All rights reserved.
A method to find an optimal draw solute for costeffective FO(forward osmosis) desalination process Tae-wooKima,Young Kimb, Choamun Yunc,Hong Janga, Woohyun Kima, Sunwon Parka* a
Department of Chemical and Biomolecular Engineering, Korea Advanced Institute of Science and Technology, 373-1 Guseong-dong, Yuseong-gu, Daejeon,305-701, South Korea b Korea Institute of Machinery and Materials, 156 Gajeongbuk-Ro, Yuseong-gu, Daejeon, 305-343, South Korea c Fuel Cell System Development Team, Corporate R&D Institute, Doosan Heavy Industries & Construction, 463-1, Jeonmin-Dong, Yuseong-gu, Daejeon, 305-811, South Korea
Abstract A method to find a draw solute for forward osmosis desalination is developed. The economics of a forward osmosis process is largely influenced by the selection of a draw solute and energy consumed for its separation from the fresh product water. The proposed method evaluates a vast number of possible draw solutes by predicting the production rate of fresh water, investment cost and energy consumption in the separation process. The possible draw solutes are confined to electrolytes and its separation processes to thermal systems. Subsequently, a cost-minimizing draw solute is determined subject to any necessary constraints. The developed method will contribute to designing commercially viable forward-osmosis desalination processes. Keywords: Draw Solute, Forward Osmosis, Desalination, Screening, Property Database
1. Introduction Water scarcity problem has been intensified in many regions all over the world by climate change, and expanding industrial and agricultural demand for water. In many of such areas of drought, desalination is often the only solution. Current commercial desalination technologies such as multi-stage flash (MSF) and reverse osmosis (RO), however, consume large thermal energy (200~400 MJ/ton in MSF process) (Borsani et al., 2005), or expensive electrical energy (4~6 kWh in RO process) (Fritzmann et al., 2007). In this regard, forward osmosis (FO) desalination is recently drawing attention to reduce the energy consumption in desalination processes. An FO desalination system consists of a membrane process and a draw solute recovery process. In the membrane process, the driving force for water permeation is the chemical potential difference between draw solution and seawater. Therefore, the osmotic pressure of draw solution should be higher than that of seawater to enable spontaneous permeation of water molecules. The draw solution diluted with the permeated water then enters the recovery process to produce fresh water while reconcentrating the solution. Note that the total energy efficiency is determined by the energy requirement of the recovery process (McGinnis et al., 2007).
A method to find an optimal draw solute for cost-effective FO(forward osmosis) desalination process
1453
The design of an FO desalination system is mainly determined with selection of a draw solute. The draw solute should have high solubility and high osmotic pressure which would lead to high water flux through the membrane. At the same time, it should be readily separable from water consuming low energy. In previous researches, several candidate compounds have been assessed for their application to desalination. The water flux at a fixed concentration is measured in those studies, often to figure out the performance of the membrane. These candidates include ammonium bicarbonate, magnesium sulfate, sodium sulfate, potassium sulfate, potassium nitrate, potassium chloride, ethanol, glucose and fructose (Cath et al., 2006) (Achilli et al., 2010). A systematic approach is proposed in this study for optimal design of a forward osmosis desalination system. This approach considers all components in the property database not to overlook any candidate of high potential. Moreover, both membrane and draw solute recovery processes are optimized for respective candidates considering water flux and energy consumption, both of which vary depending on the selection of the draw solute. In the first step, a chemical database is established to calculate solubilities and osmotic pressures. Secondly, draw solute candidates are selected by exploring the established chemical database. Lastly, the optimal desalination system is designed for each candidate draw solute and the costs of those systems are compared to find the optimal draw solute.
2. Screening method Property information of total 4,058 compounds is gathered to establish a property database from the OLI stream analyzer which is developed by OLI systems, Inc. The analyzer provides properties such as pH, ORP, viscosity, density, enthalpy based on property models and experimental data. Subsequently, the compounds are screened in hierarchical order. The screening criteria are shown in Fig 1. At first, components whose maximum osmotic pressures are lower than that of seawater (24 bar) are screened. Then, rare metals, radioactive substances and the compounds of low solubilities below 0.5 molal concentration are screened. Boiling points of the candidates are set to be below 100°C to find the components more volatile than water because boiling water requires large latent heat energy. Finally, toxic components whose concentrations should be less than 1 ppm in drinking water are screened.
Figure 1: Screening criteria
3. Mathematical models 3.1. FO membrane process model FO membrane consists of a selective layer and a support layer (Loeb et al., 1997). Water flux through the membrane (Jw) is proportional to the osmotic pressure difference between those of seawater and draw solution, and to the pure water permeability
T. Kim et al.
1454
coefficient (A). A can be experimentally determined. The effective osmotic pressure difference is less than that of bulk solutions(ʌds.b-ʌsw.b) due to the effect of internal concentration polarization (ICP) in the support layer and external concentration polarization (ECP) at the outer wall of selective layer. The actual water flux, therefore, should be calculated as below (Tan et al., 2008). (1) J w A^S ds ,b exp( J w K ) S sw,b exp( J w / k )` where K is solute resistivity for diffusion within the support layer, k is the mass transfer coefficient, ʌds.b is osmotic pressure of the bulk draw solution and ʌsw.b is osmotic pressure of the bulk seawater. 3.2. Draw solute recovery process model In this study, the draw solute recovery is confined only to thermal separation with a distillation column. Aspen Plus® is used to simulate distillation columns. As this process only concerns about the purity of product water, i.e. the bottom product, the column is designed as a stripping column. The minimum number of trays at the specified design pressure is calculated while the purity of bottom product is kept below the target concentration. At each tray, physical and chemical equilibrium states are assumed. The vessel height is calculated by multiplying the number of trays with the tray spacing. The vessel radius is decided from the vessel height and the flowrates of vapor and liquid.
4. System optimization 4.1. Investment cost model for the FO membrane process To assess the integrated performance of the FO membrane process and the draw solute recovery process, a cost minimization problem needs to be established. The six-tenth rule is applied to calculate investment cost of the FO membrane process because currently there is no commercial-scale FO membrane process established (Williams et al., 1947). This relationship has been found to provide reasonable results for individual pieces of equipment and for entire plants. The investment cost for the FO membrane process (Cinv.M) is calculated from Eq. (2). Cinv.M
§ Q / Jw Cref .M ¨ ¨ Area ref ©
· ¸¸ ¹
0.6
(2)
where Cref.M is the investment cost for the reference process, Arearef is the membrane area of the reference process. The term (Q/Jw) means required membrane area to meet target production. 4.2. Investment cost model for the draw solute recovery process To calculate the investment cost for the distillation column, Mulet-Corripio-Evans's method is applied (Mulet et al., 1947). The method calculates the purchase costs of the vessel, the platform, ladders, and trays from the vessel weight and size. 4.3. Cost minimization The cost minimization problem for unit mass of product water is as follows. · 1 § Cinv.DC Cinv.M (3) U U r U r min C ¨ Q © LDC
LM
M
¸ ¹
steam
steam
electricity
electricity
where C is the total production cost to produce a ton of product water. Usteam and Uelectricity are the unit utility & maintenance costs for steam and electricity. rsteam and relectricity are the required amounts of steam and electricity to produce a ton of product water, Cop.DC is the operation cost for the distillation column to produce a ton of product
A method to find an optimal draw solute for cost-effective FO(forward osmosis) desalination process
1455
water, LM is the lifespan expectancy and UM is the utility & maintenance cost coefficient for the FO membrane process which includes the costs for maintenance and operation. It is assumed that the operation cost for FO membrane process is negligible.
5. Case study The optimal system design is also influenced by several cost parameters. The variance in the operation cost is represented by the steam price change while that in the investment cost is expressed with the production capacity. Several cases are studied for Usteam value of 0.008$/MJ and 0.0001$/MJ, and Q value of 1 million imperial gallons per day (MIGD : 4,546 tons per day) and 100 MIGD. Note that the capacity of a typical RO plant is 0.1~35 MIGD and that of MSF plant is 20~200 MIGD. The maximum concentration is limited to 10m(molal concentration) to compensate the negligence of the back diffusion in this model. The target purity of product water is set to follow the drinking water guidelines. 5.1. Normal steam price(0.008$/MJ), small plant capacity(1MIGD) The best candidate in this case is 2-Butanone (methyl-ethyl-ketone : MEK) with vacuum operation of the distillation column. Although 2-Butanone offers relatively low water flux, it is selected because the energy requirement to satisfy the purity standard is less than the half of the 2nd best candidate, ammonium hydroxide. Meanwhile, ammonium hydroxide ranks higher than ammonium bicarbonate even though the carbonic acid ion provides additional osmotic pressure with low energy consumption. This is because the solubility of ammonium bicarbonate is relatively low. On the other hand, high solubility of ammonium hydroxide contributes to enhancing the water flux. 5.2. Low steam price(0.0001$/MJ), large plant capacity(100MIGD) Desalination systems are often located close to power plants. When the waste heat from the power plants can be utilized, the utility cost may be significantly reduced. This case searches for the best solution with steam price of $0.0001/MJ which is much lower than the market price. Because the plant capacity of this case is small, investment cost becomes a dominant factor to decide the optimal draw solute and operation condition. Accordingly, ammonium hydroxide, which requires the smallest membrane area, is also selected as the optimal solute with 2-butanone.
Figure 2: Optimization results.
6. Conclusions In this study, a systematic approach has been proposed to finding the optimal draw solute and process design. Draw solute candidates have been selected by a hierarchical
1456
T. Kim et al.
screening method. Models for the FO membrane process and the draw solute recovery process have been established to calculate water permeability and energy consumption. A cost function integrating operation and investment costs have enabled comparing the production costs of the draw solute candidates. The cost is minimized to find the optimal draw solute, the diluted concentration of the draw solution, required membrane area, as well as the number of trays and pressure in the distillation column. In case studies, production costs at different steam prices and plant capacity conditions are evaluated. The proposed approach provides an evaluation tool to compare draw solute candidates when new candidates are suggested. Moreover, the application is not limited to the production of potable water using distillation. Membrane distillation or nanofiltration can also be used in the recovery process. The target purity of product water can be set for agricultural or industrial purposes. Furthermore, this approach can be applied to systems for solution concentration or wastewater reuse with minor modifications.
References A. Achilli, T.Y. Cathb, A.E. Childress, 2010, Selection of inorganic-based draw solutions for forward osmosis applications, Journal of Membrane Science, 364, 233–24 R. Borsani, S. Rebagliati, 2005, Fundamentals and costing of MSF desalination plants and comparison with other technologies, Desalination, 182, 29-37 T.Y. Cath, A.E. Childress, M. Elimelech, 2006, Forward osmosis: principles, applications, and recent developments, Journal of Membrane Science, 281, 70–87 B.S. Frank, 1972, Desalination of sea water, US Patent 3,670,897 C. Fritzmann, J. Löwenberg, T. Wintgens, T. Melin, 2007, State-of-the-art of reverse osmosis desalination, Desalination 216, 1–76 D.N. Glew, 1965, Process for liquid recovery and solution concentration, US Patent 3,216,930 G.T. Gray, J.R. McCutcheon, M. Elimelech, 2006, Internal concentration polarization in forward osmosis: role of membrane orientation, Desalination, 197, 1-8 I.C. Karagiannis, P.G. Soldatos, 2008, Water desalination cost literature: review and assessment, Desalination, 223, 448–456 J.O. Kessler, C.D. Moody, 1976, Drinking water from sea water by forward osmosis, Desalination, 18, 297-306 R.E. Kravath, J.A. Davis, 1975, Desalination of seawater by direct osmosis, Desalination, 16, 151–155. S. Loeb, L. Titelman, E. Korngold, J. Freiman, 1997, Effect of porous support fabric on osmosis through a Loeb-Sourirajan type asymmetric membrane, Journal of Membrane Science, 129, 243249 P. McCormicka, J. Pellegrinoa, F. Mantovanib, G. Sarti, 2008, Water, salt, and ethanol diffusion through membranes for water recovery by forward (direct) osmosis processes, Journal of Membrane Science, 325, 467–478 J.R. McCutcheon, R.L. McGinnis, M. Elimelech, 2005, A novel ammonia-carbon dioxide forward (direct) osmosis desalination process, Desalination, 174, 1-11 R.L. McGinnis, M. Elimelech, 2007, Energy requirements of ammonia–carbon dioxide forward osmosis desalination, Desalination, 207, 370–382 A. Mulet, A.B. Corripio, L.B. Evans, 1981, Estimate costs of distillation and absorption towers via correlations, Chemical Engineering, 77–82 C.H. Tan, H.Y. Ng, 2008, Modeling of external and internal concentration polarization effect on flux behaviour of forward osmosis, Water Science & Technology: Water Supply, 8.5, 533-539 M. Turkay, I.E. Grossmann, 1998, Structural flowsheet optimization with complex investment cost functions, Computers & Chemical Engineering, 22, 673-686 R. Williams, 1947, Six-Tenths Factor Aids in Approximating Costs, Chemical Engineering, 54, 124–125
A method to find an optimal draw solute for costeffective FO(forward osmosis) desalination process Tae-wooKima,Young Kimb, Choamun Yunc,Hong Janga, Woohyun Kima, Sunwon Parka* a
Department of Chemical and Biomolecular Engineering, Korea Advanced Institute of Science and Technology, 373-1 Guseong-dong, Yuseong-gu, Daejeon,305-701, South Korea b Korea Institute of Machinery and Materials, 156 Gajeongbuk-Ro, Yuseong-gu, Daejeon, 305-343, South Korea c Fuel Cell System Development Team, Corporate R&D Institute, Doosan Heavy Industries & Construction, 463-1, Jeonmin-Dong, Yuseong-gu, Daejeon, 305-811, South Korea
Abstract A method to find a draw solute for forward osmosis desalination is developed. The economics of a forward osmosis process is largely influenced by the selection of a draw solute and energy consumed for its separation from the fresh product water. The proposed method evaluates a vast number of possible draw solutes by predicting the production rate of fresh water, investment cost and energy consumption in the separation process. The possible draw solutes are confined to electrolytes and its separation processes to thermal systems. Subsequently, a cost-minimizing draw solute is determined subject to any necessary constraints. The developed method will contribute to designing commercially viable forward-osmosis desalination processes. Keywords: Draw Solute, Forward Osmosis, Desalination, Screening, Property Database
1. Introduction Water scarcity problem has been intensified in many regions all over the world by climate change, and expanding industrial and agricultural demand for water. In many of such areas of drought, desalination is often the only solution. Current commercial desalination technologies such as multi-stage flash (MSF) and reverse osmosis (RO), however, consume large thermal energy (200~400 MJ/ton in MSF process) (Borsani et al., 2005), or expensive electrical energy (4~6 kWh in RO process) (Fritzmann et al., 2007). In this regard, forward osmosis (FO) desalination is recently drawing attention to reduce the energy consumption in desalination processes. An FO desalination system consists of a membrane process and a draw solute recovery process. In the membrane process, the driving force for water permeation is the chemical potential difference between draw solution and seawater. Therefore, the osmotic pressure of draw solution should be higher than that of seawater to enable spontaneous permeation of water molecules. The draw solution diluted with the permeated water then enters the recovery process to produce fresh water while reconcentrating the solution. Note that the total energy efficiency is determined by the energy requirement of the recovery process (McGinnis et al., 2007).
A method to find an optimal draw solute for cost-effective FO(forward osmosis) desalination process
1453
The design of an FO desalination system is mainly determined with selection of a draw solute. The draw solute should have high solubility and high osmotic pressure which would lead to high water flux through the membrane. At the same time, it should be readily separable from water consuming low energy. In previous researches, several candidate compounds have been assessed for their application to desalination. The water flux at a fixed concentration is measured in those studies, often to figure out the performance of the membrane. These candidates include ammonium bicarbonate, magnesium sulfate, sodium sulfate, potassium sulfate, potassium nitrate, potassium chloride, ethanol, glucose and fructose (Cath et al., 2006) (Achilli et al., 2010). A systematic approach is proposed in this study for optimal design of a forward osmosis desalination system. This approach considers all components in the property database not to overlook any candidate of high potential. Moreover, both membrane and draw solute recovery processes are optimized for respective candidates considering water flux and energy consumption, both of which vary depending on the selection of the draw solute. In the first step, a chemical database is established to calculate solubilities and osmotic pressures. Secondly, draw solute candidates are selected by exploring the established chemical database. Lastly, the optimal desalination system is designed for each candidate draw solute and the costs of those systems are compared to find the optimal draw solute.
2. Screening method Property information of total 4,058 compounds is gathered to establish a property database from the OLI stream analyzer which is developed by OLI systems, Inc. The analyzer provides properties such as pH, ORP, viscosity, density, enthalpy based on property models and experimental data. Subsequently, the compounds are screened in hierarchical order. The screening criteria are shown in Fig 1. At first, components whose maximum osmotic pressures are lower than that of seawater (24 bar) are screened. Then, rare metals, radioactive substances and the compounds of low solubilities below 0.5 molal concentration are screened. Boiling points of the candidates are set to be below 100°C to find the components more volatile than water because boiling water requires large latent heat energy. Finally, toxic components whose concentrations should be less than 1 ppm in drinking water are screened.
Figure 1: Screening criteria
3. Mathematical models 3.1. FO membrane process model FO membrane consists of a selective layer and a support layer (Loeb et al., 1997). Water flux through the membrane (Jw) is proportional to the osmotic pressure difference between those of seawater and draw solution, and to the pure water permeability
T. Kim et al.
1454
coefficient (A). A can be experimentally determined. The effective osmotic pressure difference is less than that of bulk solutions(ʌds.b-ʌsw.b) due to the effect of internal concentration polarization (ICP) in the support layer and external concentration polarization (ECP) at the outer wall of selective layer. The actual water flux, therefore, should be calculated as below (Tan et al., 2008). (1) J w A^S ds ,b exp( J w K ) S sw,b exp( J w / k )` where K is solute resistivity for diffusion within the support layer, k is the mass transfer coefficient, ʌds.b is osmotic pressure of the bulk draw solution and ʌsw.b is osmotic pressure of the bulk seawater. 3.2. Draw solute recovery process model In this study, the draw solute recovery is confined only to thermal separation with a distillation column. Aspen Plus® is used to simulate distillation columns. As this process only concerns about the purity of product water, i.e. the bottom product, the column is designed as a stripping column. The minimum number of trays at the specified design pressure is calculated while the purity of bottom product is kept below the target concentration. At each tray, physical and chemical equilibrium states are assumed. The vessel height is calculated by multiplying the number of trays with the tray spacing. The vessel radius is decided from the vessel height and the flowrates of vapor and liquid.
4. System optimization 4.1. Investment cost model for the FO membrane process To assess the integrated performance of the FO membrane process and the draw solute recovery process, a cost minimization problem needs to be established. The six-tenth rule is applied to calculate investment cost of the FO membrane process because currently there is no commercial-scale FO membrane process established (Williams et al., 1947). This relationship has been found to provide reasonable results for individual pieces of equipment and for entire plants. The investment cost for the FO membrane process (Cinv.M) is calculated from Eq. (2). Cinv.M
§ Q / Jw Cref .M ¨ ¨ Area ref ©
· ¸¸ ¹
0.6
(2)
where Cref.M is the investment cost for the reference process, Arearef is the membrane area of the reference process. The term (Q/Jw) means required membrane area to meet target production. 4.2. Investment cost model for the draw solute recovery process To calculate the investment cost for the distillation column, Mulet-Corripio-Evans's method is applied (Mulet et al., 1947). The method calculates the purchase costs of the vessel, the platform, ladders, and trays from the vessel weight and size. 4.3. Cost minimization The cost minimization problem for unit mass of product water is as follows. · 1 § Cinv.DC Cinv.M (3) U U r U r min C ¨ Q © LDC
LM
M
¸ ¹
steam
steam
electricity
electricity
where C is the total production cost to produce a ton of product water. Usteam and Uelectricity are the unit utility & maintenance costs for steam and electricity. rsteam and relectricity are the required amounts of steam and electricity to produce a ton of product water, Cop.DC is the operation cost for the distillation column to produce a ton of product
A method to find an optimal draw solute for cost-effective FO(forward osmosis) desalination process
1455
water, LM is the lifespan expectancy and UM is the utility & maintenance cost coefficient for the FO membrane process which includes the costs for maintenance and operation. It is assumed that the operation cost for FO membrane process is negligible.
5. Case study The optimal system design is also influenced by several cost parameters. The variance in the operation cost is represented by the steam price change while that in the investment cost is expressed with the production capacity. Several cases are studied for Usteam value of 0.008$/MJ and 0.0001$/MJ, and Q value of 1 million imperial gallons per day (MIGD : 4,546 tons per day) and 100 MIGD. Note that the capacity of a typical RO plant is 0.1~35 MIGD and that of MSF plant is 20~200 MIGD. The maximum concentration is limited to 10m(molal concentration) to compensate the negligence of the back diffusion in this model. The target purity of product water is set to follow the drinking water guidelines. 5.1. Normal steam price(0.008$/MJ), small plant capacity(1MIGD) The best candidate in this case is 2-Butanone (methyl-ethyl-ketone : MEK) with vacuum operation of the distillation column. Although 2-Butanone offers relatively low water flux, it is selected because the energy requirement to satisfy the purity standard is less than the half of the 2nd best candidate, ammonium hydroxide. Meanwhile, ammonium hydroxide ranks higher than ammonium bicarbonate even though the carbonic acid ion provides additional osmotic pressure with low energy consumption. This is because the solubility of ammonium bicarbonate is relatively low. On the other hand, high solubility of ammonium hydroxide contributes to enhancing the water flux. 5.2. Low steam price(0.0001$/MJ), large plant capacity(100MIGD) Desalination systems are often located close to power plants. When the waste heat from the power plants can be utilized, the utility cost may be significantly reduced. This case searches for the best solution with steam price of $0.0001/MJ which is much lower than the market price. Because the plant capacity of this case is small, investment cost becomes a dominant factor to decide the optimal draw solute and operation condition. Accordingly, ammonium hydroxide, which requires the smallest membrane area, is also selected as the optimal solute with 2-butanone.
Figure 2: Optimization results.
6. Conclusions In this study, a systematic approach has been proposed to finding the optimal draw solute and process design. Draw solute candidates have been selected by a hierarchical
1456
T. Kim et al.
screening method. Models for the FO membrane process and the draw solute recovery process have been established to calculate water permeability and energy consumption. A cost function integrating operation and investment costs have enabled comparing the production costs of the draw solute candidates. The cost is minimized to find the optimal draw solute, the diluted concentration of the draw solution, required membrane area, as well as the number of trays and pressure in the distillation column. In case studies, production costs at different steam prices and plant capacity conditions are evaluated. The proposed approach provides an evaluation tool to compare draw solute candidates when new candidates are suggested. Moreover, the application is not limited to the production of potable water using distillation. Membrane distillation or nanofiltration can also be used in the recovery process. The target purity of product water can be set for agricultural or industrial purposes. Furthermore, this approach can be applied to systems for solution concentration or wastewater reuse with minor modifications.
References A. Achilli, T.Y. Cathb, A.E. Childress, 2010, Selection of inorganic-based draw solutions for forward osmosis applications, Journal of Membrane Science, 364, 233–24 R. Borsani, S. Rebagliati, 2005, Fundamentals and costing of MSF desalination plants and comparison with other technologies, Desalination, 182, 29-37 T.Y. Cath, A.E. Childress, M. Elimelech, 2006, Forward osmosis: principles, applications, and recent developments, Journal of Membrane Science, 281, 70–87 B.S. Frank, 1972, Desalination of sea water, US Patent 3,670,897 C. Fritzmann, J. Löwenberg, T. Wintgens, T. Melin, 2007, State-of-the-art of reverse osmosis desalination, Desalination 216, 1–76 D.N. Glew, 1965, Process for liquid recovery and solution concentration, US Patent 3,216,930 G.T. Gray, J.R. McCutcheon, M. Elimelech, 2006, Internal concentration polarization in forward osmosis: role of membrane orientation, Desalination, 197, 1-8 I.C. Karagiannis, P.G. Soldatos, 2008, Water desalination cost literature: review and assessment, Desalination, 223, 448–456 J.O. Kessler, C.D. Moody, 1976, Drinking water from sea water by forward osmosis, Desalination, 18, 297-306 R.E. Kravath, J.A. Davis, 1975, Desalination of seawater by direct osmosis, Desalination, 16, 151–155. S. Loeb, L. Titelman, E. Korngold, J. Freiman, 1997, Effect of porous support fabric on osmosis through a Loeb-Sourirajan type asymmetric membrane, Journal of Membrane Science, 129, 243249 P. McCormicka, J. Pellegrinoa, F. Mantovanib, G. Sarti, 2008, Water, salt, and ethanol diffusion through membranes for water recovery by forward (direct) osmosis processes, Journal of Membrane Science, 325, 467–478 J.R. McCutcheon, R.L. McGinnis, M. Elimelech, 2005, A novel ammonia-carbon dioxide forward (direct) osmosis desalination process, Desalination, 174, 1-11 R.L. McGinnis, M. Elimelech, 2007, Energy requirements of ammonia–carbon dioxide forward osmosis desalination, Desalination, 207, 370–382 A. Mulet, A.B. Corripio, L.B. Evans, 1981, Estimate costs of distillation and absorption towers via correlations, Chemical Engineering, 77–82 C.H. Tan, H.Y. Ng, 2008, Modeling of external and internal concentration polarization effect on flux behaviour of forward osmosis, Water Science & Technology: Water Supply, 8.5, 533-539 M. Turkay, I.E. Grossmann, 1998, Structural flowsheet optimization with complex investment cost functions, Computers & Chemical Engineering, 22, 673-686 R. Williams, 1947, Six-Tenths Factor Aids in Approximating Costs, Chemical Engineering, 54, 124–125