- Email: [email protected]
distillation process
European Symposium on Computer Aided Process Engineering - 13 A. Kraslawski and I. Turunen (Editors) © 2003 Elsevier Science B.V. All rights reserved.
743
Modelling and Simulation of a Combined Membrane/Distillation Process Peter Kreis, Andrzej Gorak University of Dortmund, Chemical Engineering Department, Dortmund, Germany Email: [email protected]. Fax: +49 231 755-3035
Abstract This theoretical study is focused on the process combination of a distillation column and a pervaporation unit located in the side stream of the column. This hybrid membrane process can be applied for the separation of azeotropic mixtures such as acetone, isopropanol and water. Water is removed from the side stream of the column by pervaporation, while pure acetone and isopropanol are obtained at the top and bottom of the column. Detailed simulation studies show the influence of decisive structural parameters like side stream rate and recycle position as well as operational parameters like reflux ratio and mass flow on concentration profiles, membrane area and product compositions.
1. Introduction Distillation is still the most common unit operation to separate liquid mixtures in chemical and petroleum industry because the treatment of large product streams and high purities with a simple process design is possible. Despite of this the separation of azeotropic mixtures into pure components requires complex distillation steps and/or the use of an entrainer. Industrial applied processes are azeotropic, extractive or pressure swing distillation (Stichlmair and Fair, 1998). Another sophisticated method for the separation of binary or multicomponent azeotropic mixtures is the hybrid membrane process, consisting of a distillation column and a membrane unit. The effects of synergy for such an integrated process were investigated in recent theoretical studies, e.g. for the dehydration of alcohols (Sommer et al., 2002), for the production of fuel additives such as MTBE (Hommerich and Rautenbach, 1998b) or for the separation of non-ideal ternary alcohol/water mixtures (Kuppinger et al., 2000), (Brusis et. al, 2001). Process integration allows for significant reduction of equipment and operational costs as well as considerable energy saving compared to conventional distillation processes. Despite of all these advantages membrane separation is not yet established in chemical industry due to low permeate fluxes, short membrane lifetime or the lack of general design methodology and detailed process know-how. As recent studies show a promising progress in development of reliable high flux membranes, it is very likely that such hybrid processes will be applied in industrial scale in the near future. In this work the separation of the ternary mixture of acetone, isopropanol and water using a hybrid membrane process is studied. This non-ideal mixture with a minimum-
744 boiling azeotrop between isopropanol and water occures in the production of acetone via isopropanol (Turton et al., 1998).
FeedEZZzN
Sweep gas I
!^
PV:
VP:
Pfeed>Psa.
Pfeed ^Psa
+4-+
L l Z : ^ Retentate
I
j> Permeate
Figure 1: Principle ofpervaporation (PV) and vapour permeation (VP).
2. Pervaporation Besides high selectivity and compact design, pervaporation (PV) and vapour permeation (VP) facilitate the simple integration into existing processes. Therefore both membrane processes are very suitable for hybrid processes. The principles of pervaporation and vapour permeation are very similar. Volatile components are separated by a non-porous membrane due to different sorption and diffusion behaviour. Consequently the separation is not limited by the vapour-liquid equilibrium which is the main advantage as compared to common mass transfer processes. The driving force is the gradient of the chemical potential which is generated by lowering the partial pressure of the most permeating component on the permeate side. Usually this is achieved by applying vacuum and/or an inert sweeping gas. The main difference between PV and VP is that the feed in VP is supplied as vapour whereas in PV the feed components change their aggregate state from liquid to vapour while permeating through the membrane. The energy to vaporise the permeate is provided by the liquid feed stream. Therefore the liquid stream exits the membrane module at a decreased temperature. A characteristic parameter of membrane processes is the permeability. In general the permeability P, is proportional to diffusivity DiMemb and solubility Si^Memb of each component in the membrane material: M,Memb ~
•^i,Memb ' ^i,Memb
v^)
The parameters and consequently the efficiency of PV strongly depends on the properties of the membrane material. Common membrane materials are various dense polymers and microporous inorganic membranes (zeolithes, silica, ...) either with hydrophilic or organophilic character. Furthermore composite membranes offer the possibility to combine different materials for the dense active layer and the porous support layer. Besides membrane material fluid hydrodynamics influences the efficiency of separation. The pressure drop especially on the permeate side reduces the driving force of the most permeating components.
745
3. Hybrid Membrane/Distillation Processes Depending on thermodynamic properties of the mixture the hybrid process offers multiple configuration options in order to combine the membrane module and the distillation column. Large number of separation stages and high reflux ratios are necessary to fractionate close boiling components using conventional distillation processes. For separation of such mixtures the membrane is located in the side stream of the column (fig. 2a). Both streams, permeate and retentate are fed back to the column. Due to higher separation efficiency the membrane assists the separation in the column. This leads to a significant reduction of column stages.
JT
^
f ^
"c
^
Figure 2: Hybrid membrane process to separate a) close boiling, b) binary azeotropic and c) multicomponent mixtures (Hommerich, 1998a). Most investigations are focused on the separation of non-ideal binary mixtures., e.g purification of ethanol or isopropanol. The main purpose of the membrane unit is to overcome the azeotropic point of the top product (fig. 2b). A further enrichment up to the desired product purity can either be achieved with the membrane unit or with a second column. The objective of this study is to investigate the process configuration illustrated in figure 2c. Therefore the dehydration of the ternary mixture acetone, isopropanol and water into pure components in one distillation column combined with a hydrophilic membrane unit located in the side stream of the column is analysed. The water-depleted retentate from the permeation zone is returned back to the column while the permeate is removed out of the process. In this configuration, the operation conditions for the membrane separation is more suitable because the side stream can be placed near the maximum concentration of the most permeating component which leads to an increased driving force and consequently to smaller membrane areas.
4. Modelling For a fundamental understanding of the hybrid process it is necessary to describe the interactions between two different unit operations with appropriate models. Making basic parameter studies the equilibrium stage model for distillation and a short-cut model for membrane separation is sufficient. The models are well established and the model parameters are quite accessible. This combination gives an first survey on the influence of structural and operational parameters on the concentration profiles in the column and on the maximum amount of water, which can be removed.
746
Figure 3: Hybrid process in simulation environment Aspen Custom ModelerTM On the other hand the definition of a feasible operating region using short-cut models is not possible. The prediction of the mass transfer in membranes is the decisive factor of the entire hybrid process. The resulting permeate fluxes and consequently the membrane area are very important parameters to estimate the economical potential and the feasibility of the entire hybrid process. Therefore detailed models for the membrane unit with an semi-empirical and physical background are developed in this work to characterise the membrane separation step. The flexible model structure enables the choice of different modelling approaches for permeabilities. Among them a short-cut approach with constant permeabilities of each component, a temperature dependency of permeabilities represented by the Arrhenius equation and extended model approaches (Hommerich, 1997), (Meyer-Blumenroth, 1996) are implemented to utilise different membrane materials, e.g. inorganic zeolithes or glassy and swelling polymeric membranes. Feed and permeate pressure drop, temperature loss due to permeate vaporisation and phenomena like concentration and temperature polarisation can be taken into account. Additionally different configurations like lumen and shell feed or co and counter current flow are possible. Furthermore, a rate based model for distillation (Kloker et al., 2002) is used to perform detailed process studies of the integrated process. The relevant models for the distillation column, membrane separation and peripherals are implemented into the simulation environment Aspen Custom Modeler'^'^ (fig. 3).
5. Simulation Studies The following assumptions for the theoretical studies with the developed model are made: the column diameter is 50 mm and the column is equipped with 5 meter of the structured packing SulzerBX. The feed contains 14.1 weight percent of water, 8.4 weight percent of isopropanol and 77.5 weight percent of acetone. The feed enters the column at 3 m with a mass flow of 2 kg/h. The approach of Meyer-Blumenroth is chosen to take into account the swelling behaviour of the PVA membrane material. Pressure drop on lumen and shell side is considered. The necessary model parameters were determined in lab-scale pervaporation experiments. Figure 4 illustrates the strong
747 influence of the distillate to feed ratio on the concentration profiles in the column. The reboiler heat duty is 1200 Watt and the mass flow of the side stream is set to 4 kg/h. At low distillate to feed ratios (fig. 4, left), almost pure acetone is present in the distillate, however the amount of acetone in the side stream is rather large. Therefore permeate fluxes are small and the membrane area is not sufficient to remove the total amount of feed water entering the column. If the distillate to feed ratio is increased the mole fraction of acetone in the side stream can be decreased significantly (fig. 4, right). The mole fraction of water in the side stream is high enough and the membrane enables to remove almost the total amount of feed water. Figure 5 shows the influence of heat duty and side stream mass flow on the required membrane area for the removal of 97.5% of the water entering the column. The distillate flow is 1.56 kg/h. The reference membrane area is marked in the diagram. The operational parameters are taken from the conditions mentioned above. Side stream is set to 4 kg/h and the reboiler heat duty is 1200 Watt. With increasing heat duty the suitable operation region of the hybrid process increases because high reflux ratios improve the entire separation and the liquid and vapour load in the column is increased. The water concentration in the stripping section and in the side stream is shifted to higher mole fractions. This leads to higher transmembrane fluxes and consequently smaller membrane areas. The mass flow of the side stream strongly influences the required membrane area. Liquid column profile
-0.5
0.5
1.5 2.6 3.6 Column height [m]
Liquid column profile
4.6
-0.5
0.5
1.5 2.6 3.6 Column height [m]
4.6
Figure 4: Liquid column profile of distillation at different D/F ratios. By increasing the side stream mass flow the concentration of water in the membrane feed and the water concentration difference between membrane feed and retentate is generally decreasing. In the case swelling membrane materials like PVA are applied, it is crucial that at low water concentrations the swelling of the membrane and consequently the membrane flux decrease significantly. At moderate mass flows (approx. 3-4 kg/h) small membrane areas are sufficient to reach the desired water removal. At low side stream rates the average permeate fluxes in the module are increasing but if the mass flow is raised further lower average permeate fluxes are obtained due to the phenomena described above.
6. Conclusions A flexible and robust model of pervaporation and vapour permeation with different modelling depths was developed in the simulation environment Aspen Custom
748 Modeler^^^^ Lab-scale experiments are performed to determine the model parameters of membrane separation.
spec, membrane area [m^/m^] ,1.5
2 ° 2.6 3.0 3.6 4.0
4.5 5.0 Side stream [kg/h]
Figure 5: Required membrane area to remove 97.5% of the water entering the column. The membrane model is able to describe the mass transfer through membranes and takes into account the specific effects of different membrane materials. Simulation studies with the non-equilibrium model for distillation and the semi-empirical membrane model illustrate the influence of the mass flow of the side stream and the heating energy on the required membrane area. Both parameters have a major effect on the membrane area. Rigorous models for both unit operations are necessary to perform detailed process studies of the integrated process, because all physical effects have to be taken into account especially for membrane separation.
7. References Brusis, D., Stichlmair, J. and Kuppinger, F.F., (2001), Chemie Ingenieur Technik, 73, 624. Hommerich, U., (1998a), Ph.D. Thesis, RWTH Aachen, Germany Hommerich, U. and Rautenbach, R. (1998b), J. of Membrane Science, 146, 53-64. Kloeker, M., Kenig, E.Y., Gorak, A., Markusse, P., Kwant, G., Goetze, L. and Moritz, P. (2002), In Proc. Int. Conf. "Distillation and Absorption", Baden-Baden, Germany. Kuppinger, F.-F., Meier, R. and Dussel, R. (2000) Chemie Ingenieur Technik, 72, 333338. Meyer-Blumenroth, U. (1989), Ph.D. Thesis, RWTH Aachen, Germany. Sommer, S., Klinkhammer, B. and Melin, T. (2002), Desalination, 149, 15-21 Stichlmair, J. and Fair, J.R. (1998), Distillation-Principles and Practice, Wiley-VCH, New York, 1998 Turton, R., Bailie, R.C., Whiting, W.B. and Shaeiwitz, J.A. (1998) Analysis, Synthesis and Design of Chemical Processes, Prentice Hall PTR, New Jersey.
8. Acknowledgement We are greatful to Max-Buchner Forschungsstiftung of the DECHEMA for the financial support of this research.
743
Modelling and Simulation of a Combined Membrane/Distillation Process Peter Kreis, Andrzej Gorak University of Dortmund, Chemical Engineering Department, Dortmund, Germany Email: [email protected]. Fax: +49 231 755-3035
Abstract This theoretical study is focused on the process combination of a distillation column and a pervaporation unit located in the side stream of the column. This hybrid membrane process can be applied for the separation of azeotropic mixtures such as acetone, isopropanol and water. Water is removed from the side stream of the column by pervaporation, while pure acetone and isopropanol are obtained at the top and bottom of the column. Detailed simulation studies show the influence of decisive structural parameters like side stream rate and recycle position as well as operational parameters like reflux ratio and mass flow on concentration profiles, membrane area and product compositions.
1. Introduction Distillation is still the most common unit operation to separate liquid mixtures in chemical and petroleum industry because the treatment of large product streams and high purities with a simple process design is possible. Despite of this the separation of azeotropic mixtures into pure components requires complex distillation steps and/or the use of an entrainer. Industrial applied processes are azeotropic, extractive or pressure swing distillation (Stichlmair and Fair, 1998). Another sophisticated method for the separation of binary or multicomponent azeotropic mixtures is the hybrid membrane process, consisting of a distillation column and a membrane unit. The effects of synergy for such an integrated process were investigated in recent theoretical studies, e.g. for the dehydration of alcohols (Sommer et al., 2002), for the production of fuel additives such as MTBE (Hommerich and Rautenbach, 1998b) or for the separation of non-ideal ternary alcohol/water mixtures (Kuppinger et al., 2000), (Brusis et. al, 2001). Process integration allows for significant reduction of equipment and operational costs as well as considerable energy saving compared to conventional distillation processes. Despite of all these advantages membrane separation is not yet established in chemical industry due to low permeate fluxes, short membrane lifetime or the lack of general design methodology and detailed process know-how. As recent studies show a promising progress in development of reliable high flux membranes, it is very likely that such hybrid processes will be applied in industrial scale in the near future. In this work the separation of the ternary mixture of acetone, isopropanol and water using a hybrid membrane process is studied. This non-ideal mixture with a minimum-
744 boiling azeotrop between isopropanol and water occures in the production of acetone via isopropanol (Turton et al., 1998).
FeedEZZzN
Sweep gas I
!^
PV:
VP:
Pfeed>Psa.
Pfeed ^Psa
+4-+
L l Z : ^ Retentate
I
j> Permeate
Figure 1: Principle ofpervaporation (PV) and vapour permeation (VP).
2. Pervaporation Besides high selectivity and compact design, pervaporation (PV) and vapour permeation (VP) facilitate the simple integration into existing processes. Therefore both membrane processes are very suitable for hybrid processes. The principles of pervaporation and vapour permeation are very similar. Volatile components are separated by a non-porous membrane due to different sorption and diffusion behaviour. Consequently the separation is not limited by the vapour-liquid equilibrium which is the main advantage as compared to common mass transfer processes. The driving force is the gradient of the chemical potential which is generated by lowering the partial pressure of the most permeating component on the permeate side. Usually this is achieved by applying vacuum and/or an inert sweeping gas. The main difference between PV and VP is that the feed in VP is supplied as vapour whereas in PV the feed components change their aggregate state from liquid to vapour while permeating through the membrane. The energy to vaporise the permeate is provided by the liquid feed stream. Therefore the liquid stream exits the membrane module at a decreased temperature. A characteristic parameter of membrane processes is the permeability. In general the permeability P, is proportional to diffusivity DiMemb and solubility Si^Memb of each component in the membrane material: M,Memb ~
•^i,Memb ' ^i,Memb
v^)
The parameters and consequently the efficiency of PV strongly depends on the properties of the membrane material. Common membrane materials are various dense polymers and microporous inorganic membranes (zeolithes, silica, ...) either with hydrophilic or organophilic character. Furthermore composite membranes offer the possibility to combine different materials for the dense active layer and the porous support layer. Besides membrane material fluid hydrodynamics influences the efficiency of separation. The pressure drop especially on the permeate side reduces the driving force of the most permeating components.
745
3. Hybrid Membrane/Distillation Processes Depending on thermodynamic properties of the mixture the hybrid process offers multiple configuration options in order to combine the membrane module and the distillation column. Large number of separation stages and high reflux ratios are necessary to fractionate close boiling components using conventional distillation processes. For separation of such mixtures the membrane is located in the side stream of the column (fig. 2a). Both streams, permeate and retentate are fed back to the column. Due to higher separation efficiency the membrane assists the separation in the column. This leads to a significant reduction of column stages.
JT
^
f ^
"c
^
Figure 2: Hybrid membrane process to separate a) close boiling, b) binary azeotropic and c) multicomponent mixtures (Hommerich, 1998a). Most investigations are focused on the separation of non-ideal binary mixtures., e.g purification of ethanol or isopropanol. The main purpose of the membrane unit is to overcome the azeotropic point of the top product (fig. 2b). A further enrichment up to the desired product purity can either be achieved with the membrane unit or with a second column. The objective of this study is to investigate the process configuration illustrated in figure 2c. Therefore the dehydration of the ternary mixture acetone, isopropanol and water into pure components in one distillation column combined with a hydrophilic membrane unit located in the side stream of the column is analysed. The water-depleted retentate from the permeation zone is returned back to the column while the permeate is removed out of the process. In this configuration, the operation conditions for the membrane separation is more suitable because the side stream can be placed near the maximum concentration of the most permeating component which leads to an increased driving force and consequently to smaller membrane areas.
4. Modelling For a fundamental understanding of the hybrid process it is necessary to describe the interactions between two different unit operations with appropriate models. Making basic parameter studies the equilibrium stage model for distillation and a short-cut model for membrane separation is sufficient. The models are well established and the model parameters are quite accessible. This combination gives an first survey on the influence of structural and operational parameters on the concentration profiles in the column and on the maximum amount of water, which can be removed.
746
Figure 3: Hybrid process in simulation environment Aspen Custom ModelerTM On the other hand the definition of a feasible operating region using short-cut models is not possible. The prediction of the mass transfer in membranes is the decisive factor of the entire hybrid process. The resulting permeate fluxes and consequently the membrane area are very important parameters to estimate the economical potential and the feasibility of the entire hybrid process. Therefore detailed models for the membrane unit with an semi-empirical and physical background are developed in this work to characterise the membrane separation step. The flexible model structure enables the choice of different modelling approaches for permeabilities. Among them a short-cut approach with constant permeabilities of each component, a temperature dependency of permeabilities represented by the Arrhenius equation and extended model approaches (Hommerich, 1997), (Meyer-Blumenroth, 1996) are implemented to utilise different membrane materials, e.g. inorganic zeolithes or glassy and swelling polymeric membranes. Feed and permeate pressure drop, temperature loss due to permeate vaporisation and phenomena like concentration and temperature polarisation can be taken into account. Additionally different configurations like lumen and shell feed or co and counter current flow are possible. Furthermore, a rate based model for distillation (Kloker et al., 2002) is used to perform detailed process studies of the integrated process. The relevant models for the distillation column, membrane separation and peripherals are implemented into the simulation environment Aspen Custom Modeler'^'^ (fig. 3).
5. Simulation Studies The following assumptions for the theoretical studies with the developed model are made: the column diameter is 50 mm and the column is equipped with 5 meter of the structured packing SulzerBX. The feed contains 14.1 weight percent of water, 8.4 weight percent of isopropanol and 77.5 weight percent of acetone. The feed enters the column at 3 m with a mass flow of 2 kg/h. The approach of Meyer-Blumenroth is chosen to take into account the swelling behaviour of the PVA membrane material. Pressure drop on lumen and shell side is considered. The necessary model parameters were determined in lab-scale pervaporation experiments. Figure 4 illustrates the strong
747 influence of the distillate to feed ratio on the concentration profiles in the column. The reboiler heat duty is 1200 Watt and the mass flow of the side stream is set to 4 kg/h. At low distillate to feed ratios (fig. 4, left), almost pure acetone is present in the distillate, however the amount of acetone in the side stream is rather large. Therefore permeate fluxes are small and the membrane area is not sufficient to remove the total amount of feed water entering the column. If the distillate to feed ratio is increased the mole fraction of acetone in the side stream can be decreased significantly (fig. 4, right). The mole fraction of water in the side stream is high enough and the membrane enables to remove almost the total amount of feed water. Figure 5 shows the influence of heat duty and side stream mass flow on the required membrane area for the removal of 97.5% of the water entering the column. The distillate flow is 1.56 kg/h. The reference membrane area is marked in the diagram. The operational parameters are taken from the conditions mentioned above. Side stream is set to 4 kg/h and the reboiler heat duty is 1200 Watt. With increasing heat duty the suitable operation region of the hybrid process increases because high reflux ratios improve the entire separation and the liquid and vapour load in the column is increased. The water concentration in the stripping section and in the side stream is shifted to higher mole fractions. This leads to higher transmembrane fluxes and consequently smaller membrane areas. The mass flow of the side stream strongly influences the required membrane area. Liquid column profile
-0.5
0.5
1.5 2.6 3.6 Column height [m]
Liquid column profile
4.6
-0.5
0.5
1.5 2.6 3.6 Column height [m]
4.6
Figure 4: Liquid column profile of distillation at different D/F ratios. By increasing the side stream mass flow the concentration of water in the membrane feed and the water concentration difference between membrane feed and retentate is generally decreasing. In the case swelling membrane materials like PVA are applied, it is crucial that at low water concentrations the swelling of the membrane and consequently the membrane flux decrease significantly. At moderate mass flows (approx. 3-4 kg/h) small membrane areas are sufficient to reach the desired water removal. At low side stream rates the average permeate fluxes in the module are increasing but if the mass flow is raised further lower average permeate fluxes are obtained due to the phenomena described above.
6. Conclusions A flexible and robust model of pervaporation and vapour permeation with different modelling depths was developed in the simulation environment Aspen Custom
748 Modeler^^^^ Lab-scale experiments are performed to determine the model parameters of membrane separation.
spec, membrane area [m^/m^] ,1.5
2 ° 2.6 3.0 3.6 4.0
4.5 5.0 Side stream [kg/h]
Figure 5: Required membrane area to remove 97.5% of the water entering the column. The membrane model is able to describe the mass transfer through membranes and takes into account the specific effects of different membrane materials. Simulation studies with the non-equilibrium model for distillation and the semi-empirical membrane model illustrate the influence of the mass flow of the side stream and the heating energy on the required membrane area. Both parameters have a major effect on the membrane area. Rigorous models for both unit operations are necessary to perform detailed process studies of the integrated process, because all physical effects have to be taken into account especially for membrane separation.
7. References Brusis, D., Stichlmair, J. and Kuppinger, F.F., (2001), Chemie Ingenieur Technik, 73, 624. Hommerich, U., (1998a), Ph.D. Thesis, RWTH Aachen, Germany Hommerich, U. and Rautenbach, R. (1998b), J. of Membrane Science, 146, 53-64. Kloeker, M., Kenig, E.Y., Gorak, A., Markusse, P., Kwant, G., Goetze, L. and Moritz, P. (2002), In Proc. Int. Conf. "Distillation and Absorption", Baden-Baden, Germany. Kuppinger, F.-F., Meier, R. and Dussel, R. (2000) Chemie Ingenieur Technik, 72, 333338. Meyer-Blumenroth, U. (1989), Ph.D. Thesis, RWTH Aachen, Germany. Sommer, S., Klinkhammer, B. and Melin, T. (2002), Desalination, 149, 15-21 Stichlmair, J. and Fair, J.R. (1998), Distillation-Principles and Practice, Wiley-VCH, New York, 1998 Turton, R., Bailie, R.C., Whiting, W.B. and Shaeiwitz, J.A. (1998) Analysis, Synthesis and Design of Chemical Processes, Prentice Hall PTR, New Jersey.
8. Acknowledgement We are greatful to Max-Buchner Forschungsstiftung of the DECHEMA for the financial support of this research.