Economic comparison between azeotropic distillation

Separation and Purification Technology 37 (2004) 33–49

Economic comparison between azeotropic distillation and different hybrid systems combining distillation with pervaporation for the dehydration of isopropanol Veerle Van Hoof∗ , Liesbet Van den Abeele, Anita Buekenhoudt, Chris Dotremont, Roger Leysen Vito (Flemish Institute for Technological Research), Process Technology, Boeretang 200, B-2400 Mol, Belgium Received 6 January 2003; received in revised form 17 July 2003; accepted 1 August 2003

Abstract An economic analysis was performed comparing different processes for the dehydration of isopropanol/water. Traditional azeotropic distillation was compared with a hybrid system consisting of distillation followed by pervaporation and a hybrid system consisting of distillation followed by pervaporation followed by a second distillation. In the hybrid combinations pervaporation with both polymeric (PERVAP® 2510, Sulzer Chemtech GmbH) and ceramic membranes (NaA type zeolite, Mitsui & Co.) was investigated. Both membranes were tested for the dehydration of isopropanol at 70 and 90 ◦ C. Based on these experimental results, calculations were performed with the PV design calculation program (RWTH Aachen) to obtain economic data on the pervaporation processes. For the simulations of the distillation processes, Aspen Plus 11.1 was used. It was found that the hybrid system distillation–pervaporation with ceramic membranes was the most interesting process from economic point of view and could lead to a saving in total costs of 49% compared to azeotropic distillation. © 2003 Elsevier B.V. All rights reserved. Keywords: Pervaporation; Distillation; Economic analysis; Hybrid process; Ceramic membranes

1. Introduction Distillation is the most widely used technique to separate a mixture of liquids, the separation being based on the difference in boiling temperature of the components. By successively heating up and condensing the mixture, the vapour will become ∗ Corresponding author. Tel.: +32-1433-5641; fax: +32-1432-1186. E-mail address: [email protected] (V. Van Hoof).

richer in the lower boiling component while the liquid will become richer in the higher boiling component. This technique as such can not be used for the separation of azeotropes as they have the same composition in the vapour and liquid phase. The separation of azeotropes traditionally occurs through a pressure-swing distillation, with the consequences of extra costs, or a third component is used as an entrainer in an extractive or an azeotropic distillation, with its unfavourable side effects [1,2]. Both

1383-5866/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.seppur.2003.08.003

34

V. Van Hoof et al. / Separation and Purification Technology 37 (2004) 33–49

Nomenclature Ccond Cdist

energy costs condenser(s) ( /h) energy costs for distillation process ( /h) Celec electricity costs ( /kW h) energy costs for heating initial feed Cfeed ( /h) heat capacity of cooling water Cp water (kJ/kg K) Cperm energy costs for permeate condensation ( /h) Cperv energy costs for pervaporation process ( /h) Creb energy costs reboiler(s) ( /h) Cret energy costs to reheat retentate streams between pervaporation modules ( /h) costs for steam ( /kg) Csteam Cstreams energy costs for heating distillation and pervaporation output streams ( /h) Ctotal total energy costs ( /h) costs for cooling water ( /kg) Cwater D–PV distillation–pervaporation (hybrid process) D–PV–D distillation–pervaporation–distillation (hybrid process) Hsteam condensation enthalpy of steam (kJ/kg) IPA isopropanol Qcond heat withdrew from the condenser(s) (kJ/h) Qfeed heat needed to heat up initial feed (kJ/h) Qperm heat withdrew for permeate condensation(kJ/h) Qreb heat added to the reboiler(s) (kJ/h) Qret heat needed to reheat retentate streams between pervaporation modules (kJ/h) Qstreams heat needed to heat up distillation and pervaporation output streams (kJ/h) Twater temperature change of cooling water

processes are very energy-intensive processes. Recently, pervaporation is introduced as an alternative to separate azeotropes. It has the advantage to be less energy consuming and more environmental friendly than distillation. As described by Lipnizki et al. [3] and recently also by Sommer et al. [4] combining distillation with pervaporation can lead to significant savings in investment and operating costs. In pervaporation the separation is not based on the relative volatility of the components in the mixture, but only depends on the relative affinity of the components for the membrane [5]. A vacuum is kept on the permeate side of the membrane while the feed side of the membrane is kept at atmospheric or elevated pressure so a pressure difference is created over the membrane which is the driving force for the pervaporation process. The component(s) that permeates through the membrane evaporates while passing through the membrane because the partial pressure of the permeating component(s) is kept lower than the equilibrium vapour pressure. There are different types of membranes for pervaporation: organic membranes, also called polymeric membranes, and inorganic membranes, also called ceramic membranes. At this moment, most pervaporation membranes that are used in industrial applications are of the polymeric type. A major drawback of these polymeric membranes is their limited solvent and temperature stability [6]. Generally, ceramic membranes are solvent and temperature stable, can be used in a broad pH range and have both high selectivity and permeability. The industrial use of ceramic membranes could lead to a higher product quality and could broaden the application range of pervaporation. The exact permeation mechanism for the pervaporation process is not yet known, however the sorption–diffusion model is the most widely accepted model to describe the pervaporation process in polymeric membranes. This model divides the pervaporation process in three consequent steps: first, there is a specific sorption in the membrane layer of one or more components in the feed mixture, depending on the characteristics of the membrane. The second step is the diffusion of the components through the membrane. According to the sorption–diffusion model, the pervaporation membrane is not porous, so diffusion is


the only possible transport mechanism. The third step is the desorption of the components on the permeate side as vapour [7]. In most applications, pervaporation is not used as a stand-alone process but it is combined with a second technique, for example a normal distillation or a chemical reactor, in a so-called hybrid process [3]. The aim of this article is to give an economic comparison between three possible processes to dehydrate a isopropanol/water mixture. The first possible process is traditional azeotropic distillation. The second possible process is a hybrid process which consists of distillation to obtain the azeotrope followed by pervaporation up to the final desired water concentration. The third possibility is a hybrid process which consists of distillation to obtain the azeotrope followed by pervaporation to break the azeotrope followed by a second distillation to dehydrate up to the final desired water concentration. For the hybrid processes, pervaporation with both polymeric or ceramic membranes is investigated to find out which membranes would be the most interesting from economic point of view. The economic analysis was performed for the dehydration of IPA. Large amounts of IPA are used in the industry for example in the electronics industry [8]. The IPA that is sold for commercial use contains between 99.0 and 99.7 wt.% IPA. Therefore, the aim of all three processes studied will be to obtain IPA with a final purity of 99.5 wt.% IPA. In a previous article Sommer et al. [4] briefly describe the economic benefits of the dehydration of isopropanol using a D–PV hybrid system with both polymeric membranes or PERVAP® SMS silica membranes and a D–PV–D hybrid system with the former silica membranes, compared to the conventional process. In the present article a more detailed economic study is performed and the ceramic membrane studied is the Mitsui NaA type zeolite membrane. For the simulation of the distillation process, ASPEN PLUS® was used. For studying the pervaporation process, the subroutine PV design, developed by the ‘Institut für Verfahrenstechnik’ RWTH Aachen, was used. This model requires the input of experimental data, therefore lab-scale pervaporation experiments were performed with both polymeric and ceramic membranes.

35

2. Materials and methods 2.1. Pervaporation experiments 2.1.1. Chemicals In each experiment a mixture of water and IPA was prepared. The IPA was obtained from Merck. The boiling point of IPA is 82 ◦ C. At a pressure of 1 bar IPA-water forms an azeotrope at 87 wt.% IPA and 13 wt.% water with a boiling point of 80 ◦ C. 2.1.2. Membranes Two hydrophilic membranes were tested for the dehydration of IPA, a polymeric membrane and a ceramic membrane. The polymeric membrane that was used was the PERVAP® 2510 membrane which was purchased from Sulzer Chemtech in Germany. This is a flat asymmetric composite membrane with a support layer of non-woven porous polyester on which an ultrafiltration membrane is cast. The selectivity can be attributed to the 0.1 m thick top layer of cross-linked PVA. This membrane is typically used for the dehydration of neutral solvents (typically isopropanol) with a maximum feed water concentration of 25%. The ceramic membrane that was used was a tubular NaA type zeolite membrane from Mitsui & Co. The membrane consists of a tubular ceramic porous carrier tube with an inner diameter of 9 mm, an outer diameter of 12 mm and a length of 200 mm which is covered on the outside with a top layer of NaA type zeolite. The support consists of a mixture of mullite, Al2 O3 and cristoballite with a Al2 O3 /SiO2 ratio of approximately 65/35%, a porosity of approximately 40% and a mean pore size of 1 m. 2.1.3. Pervaporation experiments Fig. 1 shows a schematic presentation of the pervaporation unit used to perform the pervaporation experiments. The feed tank was filled with approximately 2 l of the feed mixture, that was kept at constant temperature. A circulation pump (Speck, type SPY 2071.0671) was used to pump the feed mixture around in the pervaporation unit. The feed mixture was sent over the membrane and the retentate was sent back to the feed tank. To collect the permeate, two stainless steel vessels were used to perform continuous pervaporation. To

36


Fig. 1. Schematic diagram of lab-scale pervaporation unit.

create a vacuum on the permeate side, a vacuum pump (Pfeiffer, type DUO 004 B) was used which could be switched between the two permeate vessels. The permeate vessels were cooled with liquid nitrogen and a heating element was put around the entrance of the permeate vessels to avoid freezing of the permeate that would block the entrance of the permeate vessels. Two different pervaporation modules were used, a flat module for the Pervap® 2510 membrane, with an effective membrane area of 178 cm2 , and a tubular module for the Mitsui NaA type zeolite membrane, with an effective membrane area of 70 cm2 . Fluxes were determined by weighing the permeate collected in the permeate vessels. At each sampling time also a feed sample was collected. The total fluxes were calculated using the following formula: =

St

where J is the total permeate flux (g/h m2 ), m the mass of permeate (g), S the membrane surface area (m2 ) and t the permeation time (h). All experiments were performed at a feed flow rate between 4 and 5 l/min. With both the polymeric

and the zeolite membrane, two experiments were performed, one experiment at a feed temperature of 70 ◦ C and atmospheric feed pressure and one experiment at a feed temperature of 90 ◦ C and a feed pressure of approximately 1.6 bar. 2.1.4. Sample analyses The amount of water in the feed samples was determined by Karl Fisher titration (Metrohm, 720 KFS Titrino). The amount of water in the permeate samples was determined by analysing the amount of IPA in the samples with headspace GC/MS. GC analysis was carried out using a Hewlett-Packard type 6890 equipped with a Hewlett-Packard Voc column with a length of 30 m, an internal diameter of 0.2 mm and a film thickness of 1.8 m. 2.2. Hybrid processes To compare the different processes we assume that in each process we have to purify 1000 kg/h of a mixture containing 50 wt.% IPA and 50 wt.% water up to a final IPA concentration of 99.5 wt.%. The temperature of the initial feed mixture is 20 ◦ C. A relatively


Fig. 2. Azeotropic distillation process.

37

38


‘small’ stream of 1000 kg/h was chosen because the purification process is intended for solvent recuperation and not solvent production. The purification of IPA traditionally occurs through azeotropic distillation. Fig. 2 gives an overview of an azeotropic distillation process for the dehydration of IPA, using benzene as an entrainer. In a first column, the aim is to obtain a top product with a composition close to the azeotropic composition (we will consider distillation up to 86.5 wt.% IPA and 13.5 wt.% water). In the second column, the actual azeotropic column, an entrainer is added to break the azeotrope. The bottom stream of this column contains IPA with the final desired purity, while the top stream is a mixture of water, IPA and benzene. This top stream is sent to a decanter, where it separates in two fractions, a benzene-rich fraction and a water-rich fraction. The benzene-rich fraction is sent back to the second column, while the water-rich fraction, containing IPA, water and benzene, is sent to a third distillation column to further recover the benzene. The bottom stream of this third column is a mixture of IPA and water with almost the same composition as the initial feed mixture so it is sent back to the first column. The top stream of the third column contains water, benzene and isopropanol and is sent back to the second column. Azeotropic distillation is a very energy-consuming process and the use of an entrainer might cause unwanted impurity in the product and side streams. Therefore, other techniques should be considered to overcome these disadvantages. A very interesting alternative might be a hybrid process combining distillation with pervaporation. First the feed mixture is sent to a distillation column to obtain a top product that has a composition close to the azeotrope (we will consider distillation up to 83.0 wt.% IPA and 17.0 wt.% water). The top product of this distillation column is further dehydrated by a pervaporation unit. The top product can be dehydrated up to the final desired water concentration by pervaporation (Fig. 3) or it can be dehydrated partially by pervaporation to break the azeotrope (we will consider distillation up to approximately 5 wt.% water) and then sent to a second distillation column in which it is dehydrated up to the final desired IPA concentration (Fig. 4). Both hybrid systems will be investigated for pervaporation with polymeric membranes on the one hand and with ceramic membranes on the other.

2.3. Simulation programs To simulate the pervaporation process, the calculation program PV design, developed by the Technical University of Aachen was used. This program makes it possible to calculate the energy requirement, required membrane area, permeate flow and retentate flow for a certain pervaporation process based on experimental results obtained with the polymeric and ceramic membrane. To simulate the distillation processes, Aspen Plus® 11.1, developed by Aspentech, was used. The RadFrac model was chosen to describe the different distillation columns, and the NRTL-model was used to describe the liquid-vapor interactions. The non-random two liquid (NRTL) model is based on the real interaction parameters between the components. To reach the assumed qualities of top and bottom streams, and minimize the number of plates and reflux ratio, the design specs option in the program was used. Aspen Plus® predicted optimal values for the number of plates, the reflux ratio and the concurrent purity of the different streams. The energy consumption for the distillation processes was also calculated using Aspen Plus® .

3. Results and discussion 3.1. Pervaporation results Fig. 5 shows the total fluxes obtained with the ceramic and polymeric membrane at a feed temperature of 70 and 90 ◦ C. Fig. 6 shows the permeate water concentrations obtained with the ceramic- and polymeric membrane at a feed temperature of 70 and 90 ◦ C. Both membranes were found to be very well suited for the dehydration of IPA as both membranes show high fluxes and high selectivity. As expected both membranes showed an increase in total flux with increasing temperature. For the polymeric membrane the selectivity increased with increasing temperature while for the ceramic membrane the selectivity was very high and approximately the same at both temperatures. When the selectivity of the ceramic and polymeric membrane are compared it is clear that the ceramic membrane keeps its


39

Fig. 3. Hybrid system distillation–pervaporation with polymer membranes (a) and ceramic membranes (b).

selectivity up to very low feed water concentrations while the polymeric membranes loses its selectivity at these low feed water concentrations. From Figs. 5 and 6 it is clear that, in the range of interest (below the azeotrope, from 13 to 0 wt.% water), the ceramic membrane shows significantly higher fluxes and higher permeate water concentrations (up to 99.97 wt.% water) than the polymeric membrane.

3.2. Simulations All distillation columns were operated at a pressure of 1 bar and no pressure losses were assumed. 3.2.1. Azeotropic distillation Fig. 2 gives an overview of the results for the azeotropic distillation process. The first distillation column had 10 theoretical stages, was fed on stage 9

40
