Hybrid Systems for the Separation of Light Hydrocarbon Mixtures RALUCA ISOPESCU, CRISTIANA LUMINITA GIJIU, DANIEL DUMITRU DINCULESCU * University Politehnica of Bucharest, Department of Chemical and Biochemical Engineering, 1-7 Polizu Str., 011061, Bucharest, Romania Hybrid system and thermal coupling were investigated for the separation of light olefin/paraffin (C 2 ) at high purity. The use of ZIF-8 membrane for the separation of propene/propane mixture and the implementation of a divided wall distillation column were proposed. The implementation of such devices in the classical separation configuration scheme leads to an important reduction of energy consumption. The analysis of all separation schemes was performed using Aspen HYSYS TM v7.3 process simulator. The mathematical model for the membrane gas separation was solved using Matlab. Keywords: light hydrocarbons, membrane separation, hybrid system, divided wall distillation column Distillation is the most commonly used method for hydrocarbon mixtures separation in oil refinery and petrochemical industries, but its high energy consumption is an important disadvantage. Light olefin/paraffin mixtures (C 2 ) are the source of many row materials in petrochemical industry: ethene, propene, butane, i-butane, butenes. The separation of these compounds, at high yields and purity, is a quite difficult task, as their relative volatility is low over a large temperature range. The necessity of reducing the energy consumption led to intensive study of thermal coupling possibilities but also to finding an alternative solution for distillation, such as membrane separation. While partially or totally thermal coupling solutions are already in practical use, the membrane separation for light hydrocarbons mixture is still in the stage of membrane synthesizing and characterization. In the present work, hybrid sequences using distillation columns and membrane separation units are investigated by process modeling and simulation. Non-conventional distillation column structures, based on thermal coupling, are also studied, as part of the hybrid system, aiming to reduce at a higher extend the overall energy consumption. The efficiency of partial or totally thermal coupled distillation columns was extensively investigated in recent literature [1-4]. The divided wall distillation column (DWC) was proved to be one of the most efficient structures in terms of energy and cost savings. The DWC is the integration of two thermally coupled distillation columns into one shell (fig. 1) by providing a vertical wall that divides the distillation space into two parts: one part works as the prefractionators, where a sloppy separation is realized, and the other part represents the main column [3, 5, 1]. In the main column the light component (A) is separated as top product, the heavy component (C) is the bottom product while the middle component (B) is evacuated as side stream. This structure, which is already in practical use for some distillation operations, avoids the remixing [5] and allows the separation of the middle component at high purities, which is not possible in a simple distillation column with side-draw. From the thermodynamic point of view, a DWC is equivalent to the totally thermal coupled structure, the so called Petlyuk-configuration (figure 1), if the dividing wall is insulated, and the heat transfer through the wall is neglected [3, 6, 7]. Fig. 1. Three components mixture separation : (a) direct sequence (b) Petlyuk configuration (c) DWC Several applications for hydrocarbon or alcohol mixtures separations, extractive or azeotropic distillation [1, 3, 8-10] using the DWC are reported in literature. More than 90 applications in production scale are known and the advantages are obvious. Depending on the case considered, the energy and investment costs are reduced up to 40% compared to conventional technologies [11]. The membrane processes represent a new approach in hydrocarbon mixture separation. Membrane gas separation equipment have significantly grown in the last years [12] involving mainly non-condensable gases separation. A large potential market for membrane gas separation is the hydrocarbon mixtures separation, such as C 4 fraction [16, 17], C 2 and C 3 fraction [13]. That is why, new membrane types were investigated to cope with oil refinery needs. Zeolitic imidazole framework (ZIF-8) membranes were fabricated and first applied to gasoline vapour recovery [14]. The separation possibility of propene/propane using ZIF-8 membranes was extensively studied [13, 19] proving that the separation is mainly governed by diffusive separation. Some hybrid separation sequence, at laboratory scale, were also investigated for ethene/ethane and propene/propane separation [18] showing that membrane processes can be an alternative for energy intensive hydrocarbon separation and reducing carbon footprint in olefins production units. The present paper intends to analyze some alternative separation schemes for light hydrocarbons mixtures separation, using both membrane units and thermal coupling techniques. *email: d_dinculescu@chim.upb.ro REV. CHIM. (Bucharest) 66 No. 2 2015 http://www.revistadechimie.ro 259
Table 1 TYPICAL GAS CATALYTIC CRACKING COMPOSITION (MOL FRACTIONS) Fig. 2 Basic separation flowchart Experimental part Problem formulation A typical catalytic cracking gas composition (table 1) was considered in the analysis of separation schemes [21]. Classical separation flowcharts include, according to the direct separation sequence (the most volatile component separates first), distillation columns assigned for: methane separation as top product (de-methanizer), separation of C 2 fraction (de-ethanizer) separation of C 3 fraction (de-propanizer) and, additional, the distillation columns for the separation of ethene/ethane and propene/ propane mixtures (fig.2). The ethene/ethane separation is not difficult to be performed if the ethene purity is about 95%, as required for further organic syntheses, but becomes difficult if ethene product is aimed for polyethylene production, where a purity of 99.8% is necessary [21]. In both cases, the low temperatures required in the condenser, as well as the high pressures increase the operating costs. The propene/ propane separation is the most difficult task, as the very low relative volatility drastically affects both the minimum reflux ration and number of theoretical trays with increased propene purity demands. The results obtained for a feed of 500 kmol/h at a pressure of 35 bar, considering the direct sequence represent a base for the analysis of further improvement possibilities using membrane separation technique and thermal coupling. The analysis of all separation schemes was performed using the Aspen HYSYS TM v7.3 process simulator. In all simulated cases, the product purity constrains were defined as (mol fractions): ethene > 0.99, ethane > 0.99, propene > 0.99, propane > 0.95 and C 4 fraction > 0.98. Results and discussions Classical direct sequence separation The simulation of the classical separation sequence (fig. 2) considers some generally accepted operating conditions. The working pressures in the distillation columns were imposed in order to ensure for the top products a feasible condensing temperature. The first column (T100), where methane and hydrogen are separated as overhead gases, works at 34 bar, as mentioned in most technical reports [21] and the top temperature is 99.5 C. For the de-ethanizer (column T101) the working pressure was set at 19 bar [20], which ensures low C 3 losses in the top product. The condensing temperature of the top product is 19.7 o C that is realized by the refrigeration system. The de-propanizer (T102) 260 working conditions were set at a pressure of 18 bar corresponding to a condensing temperature of 45 o C, for which cooling water may be used. The ethene-ethane separation column is defined to produce 99.8 mol % ethene in the top. The working pressure was set to 15 bar when a condensing temperature of the top product is 38.9 C. It is well known that the ethene -ethane separation is the most refrigerant consuming tower for light hydrocarbon separation. In order to keep the reflux ratio at an acceptable value, a column with 48 theoretical trays proved to realize the required separation. According to simulation results, the distillation column assigned for propene-propane separation required a great number of theoretical trays (125 trays), and a reflux ratio of 24. Similar high values are also mentioned in literature [13, 21, 22] and technical reports. Practical cases use a series of two distillation columns building up the required number of theoretical trays [22, 23]. For all distillation columns, the number of theoretical trays was calculated in the frame of Aspen HYSYS v7.3 simulator using a short-cut design option based on Fenske- Underwood-Gilliland equations. The final structure of the columns was reached by successive simulations aiming to ensure the purity requirements with minimum energy consumption. The simulation was carried out using Peng Robinson thermodynamic model recommended for light hydrocarbon distillation. The main results are presented in table 2. Hybrid system option Recently, for gas and vapour separation, the membrane techniques are more and more used mainly for noncondensing gases: nitrogen from air; carbon dioxide from methane; and hydrogen from nitrogen, argon, or methane. The separation of condensable gases such as propene from propane is a most challenging approach as this separation is the most intensive one in petrochemical industries [13]. Recent studies proved that zeolitic-imidazolate frameworks (ZIFs) are new candidates for use in gas and vapour separation [13,14]. Among the varieties of ZIFs, ZIF-8 is the most promising candidate for propene/propane separation. Based on the results published in [19], the simulation of propene/ propane separation using ZIF-8 membrane was carried on using the mathematical model (relations (1)) implemented in Matlab: where: P total permeate rate, kmol/h; http://www.revistadechimie.ro REV. CHIM. (Bucharest) 66 No. 2 2015 (1)
Table 2 MAIN SIMULATION RESULTS FOR THE DIRECT SEPARATION SEQUENCE Fig 3. Hybrid system starting from the direct sequence separation flowchart P ma membrane permeability for δ m membrane thickness, m; A m membrane aria, m 2 ; Π ma membrane permeance for y P molar fraction of A in permeate; x R molar fraction of A in retentate; p R retentate pressure, Pa; p P permeate pressure, Pa; The membrane unit is aimed to replace the distillation column T104 in figure 2, and the separation sequence will become a hybrid one (fig. 3). The use of a membrane unit to replace the distillation column for the propene/propane separation is considered to drastically diminish the energy consumption as neither cooling nor heating utilities are necessary and the membrane feed are already pressurized. As the membrane unit is to be fed with vapour hydrocarbon mixture, the stream containing the C 3 fraction was withdrawn in vapour phase from the distillation column T102 (fig.2). The selection of a partial condenser in column T102 is convenient also from the energy consumption point of view, as it will diminish the thermal duty compared to the classical total condenser option used in the basic separation scheme (2152 kw for C 3 fraction in vapor phase compared with 2749 kw in the basic sequence). The vapour from T102 fed in membrane unit was assimilated with a binary mixture, as it contains mainly propane and propene and only 5 10-3 (mol fr.) other components. The permeate flowrate was calculated depending on the maximum permeate concentration (minimum retentate concentration) [24], in order to achieve the desired separation in one membrane unit. The main characteristics of the membrane unit and the efficiency of propenepropane separation are synthesized in table 3. The required membrane surface resulted in the specified working conditions is relatively high, about 4719 m 2, but a series of membrane module can be considered in order to optimize the space requirement. Hybrid system with a DWC Apart from replacing the propene-propane distillation column with a membrane unit, further energy reduction can be realized by using a totally thermal coupled structure, the DWC. In what concerns the thermal coupling in the separation of C 2 hydrocarbon, the main difficulties arise due to the different operating pressures. The general admitted drawback of a DWC is that it must operate at the highest working pressure of the two columns in order to ensure the condensing temperature of the distillate. A higher pressure may diminish the relative volatility compared with the classical scheme, and it will increase the bottom product boiling point compared with the classical two columns sequence, where the heavy REV. CHIM. (Bucharest) 66 No. 2 2015 http://www.revistadechimie.ro 261
Table 3 CHARACTERISTICS OF THE MEMBRANE UNIT REPLACING T104 IN THE DIRECT SEQUENCE SCHEME component is obtained in a column working at lower pressure. That is why the totally thermal coupling is generally not recommended when the difference in operating pressures between the two classical distillation columns is too large. Analyzing the flowchart (fig. 2), the thermal coupling proposed is the replacing the column T101 and T102 with a DWC to produce C 2 fraction as top product, C 3 fraction as side stream and C 4 fraction as bottom product (fig. 4). The working pressure in T100 (the desub-flowsheet (fig. 5) are automatically converged to the steady-state value during the simulation. For the thermally coupled solution, the side draw must be in vapour phase, as it will be the input in the membrane unit. The vapour side-draw may be technologically more complicated, but this option is nevertheless feasible, and control possibilities can be defined [25]. The working pressure in the DWC was set at 19 bar to cope with condensing temperature of C 2 fraction. In these conditions, a main column with 48 theoretical trays was defined and 25 trays in the pre-fractionator. The trays corresponding to the inter-connecting steams were set so that the number of trays on both sides of the dividing wall to be equal and thus ensure the same pressure drop. The variation of temperature and composition along the DWC trays is presented in figures 6-7. In these figures, the trays are represented in a continuous way. Trays numbered 1 to 25 represent the prefrationator, trays from 26 to 73 correspond to the main column. The feed tray is number 10, while side-draw tray is number 46. As figure 6 shows, the temperature on the left side of the wall (prefractionator trays 1-25) and right side (main column trays 36-60) of the DWC are very close, and in this case the assumption of neglecting the heat transfer through the wall remains valid. The variation of concentrations of the dominant components in fractions C 2 (ethane), C 3 (propene), and C 4 (1-butene), as presented in figure 7, shows an important split between ethane and butene in the pre-fractionator and a high concentration of propene in the range of trays 44-48 where the side draw is placed (tray 23 in the main column). The purities of the DWC products are: C 2 fraction 99.5% (top), C 3 fraction 99.2 % (side-draw), C 4 fraction 98.8 % (bottom). These values are very close to those realized by the sequence of T101 and T102 distillation columns (table 2) and will ensure the ethene and propene final purity demands. The C 2 fraction will be separated in the distillation column T103 while C 3 fraction is fed to the gas separation membrane unit (fig. 4) The main characteristics for the membrane gas separation unit fed with C 3 fraction obtained as side-draw in the DWC are summarized in table 4. Fig. 4 Thermally coupled separation sequence and membrane unit ethanizer) is 19 bar while the pressure in T102 (the depropanizer) is 18 bar. The pressure difference is acceptable and a DWC was tested as the use of a single condenser and a single reboiler may bring important advantages. Considering that no heat transfer occurs through the dividing wall, the DWC was modelled in HYSYS TM using the thermodynamically equivalent structure with prefractionator and main column (fig. 5) defined in a column sub-flowsheet. The number of theoretical trays in each section, the feed and side draw locations were defined starting from a short-cut design that was further slightly modified [3, 14] to reach the purity specification of the products at a low value for the energy consumption in this complex structure. The internal flows represented in the 262 Fig 5. The prefractionator and main column model used for DWC simulation http://www.revistadechimie.ro REV. CHIM. (Bucharest) 66 No. 2 2015
Fig 6. Temperature profile in the DWC In this configuration, as the membrane unit feed is at higher temperature then the previous studied case, and as the propene permeance decreases with the temperature [19], the calculated membrane surface was 3810 m 2. For industrial purposes similar membrane surfaces were considered, and reported in literature, as for instance those presented by Chilukuri P. et al. [26]. Fig 7. Composition profile in the DWC Table 4 CHARACTERISTICS OF MEMBRANE UNIT FED WITH C 3 FRACTION FROM THE DWC Energy reduction analysis by thermal coupling and membrane separation unit The comparative analysis of the different separation flowcharts was performed in terms energy consumption. The energy consumptions are synthesized in tables 5 and 6. Data in table 5 present the energy reduction realized by replacing the de-etanizer (T101) and de-propanizer (T102) with a DWC. The results in terms of energy reduction are consistent with other DWC used in hydrocarbon separations reported in literature [1, 3, 5] proving that the DWC is a good solution. As can be noticed, the reflux in the DWC is higher than the reflux in the de-ethanizer, as it must ensure the liquid flow on both sides of the dividing wall, and this is the cause of an increase of refrigeration demand with about 6 % compared with the conventional separation scheme. The energy reduction provided by the reboiler duty decrease is important while the increase of the heavy component boiling point is not significant. Data in table 6 present the energy reduction realized by the hybrid separation sequences compared to the classical direct scheme. The best solution in terms of energy reduction is the hybrid sequence with totally thermal integration of the de-ethanizer and de-propanizer building a DWC. The membrane separation unit used to replaces Table 5 MAIN RESULTS OBTAINED BY THERMAL COUPLING IN A DWC WITH VAPOUR SIDE-DRAW REV. CHIM. (Bucharest) 66 No. 2 2015 http://www.revistadechimie.ro 263
Table 6 ENERGY CONSUMPTIONS FOR THE ANALYZED SEPARATION SCHEMES one of the most energy intensive separation, propene/ propane, proved to be promising solution. A more complete analyze concerning the capital cost for such a hybrid system would validate the viability of the proposed method. This analyze is not yet possible because commercial membrane units are not available for propene-propane separation, and this is a task for further work. Conclusions The present study proved that the use of a gas separation membrane unit for propene/propane separation is possible and in what concerns the membrane surface requirement the solution is feasible is a series of ZIF-8 membrane is used. In terms of energy consumption the hybrid system proposed is very efficient. The thermal coupling technique and complex distillation columns are also efficient means for lower energy consumptions. Replacing two distillation columns, the de-ethanizer and de-propanizer from the direct separation scheme, by a DWC increases the energy reduction in the C 2 separation process. A hybrid sequence with a DWC may lead to important reduction of energy consumption, over 70%, compared with a classical separation sequence. References 1.PREMKUMAR, R., RANGAIAH, G. P., Chem. Eng. Res. Des., 87,2008, p. 47 2.RONG B.G., Chem. Eng. Res. Des., 89 (8), 2011, p. 1281 3.ISOPESCU, R., WOINAROSCHY, A., DRAGHICIU L., Rev. Chim. (Bucharest), 59, no. 7, 2008, p. 812 4.DRAGHICIU, L., ISOPESCU, R., WOINAROSCHY, A., Rev. Chim. (Bucharest), 60, no. 10, 2009, p. 1056 5. HERNANDEZ, S., SEGOVIA-HERNANDEZ, J. G., RICO-RAMIREZ, V., Energy 31, 2006, p. 2176 6.TAVAN, Y., RIAZI, S. H., NOZOHOURI, M., Energ. Convers. Manage., 79, 2014, p. 590 7.KISS, A.A., SUSZWALAK, D.J.P.C., Sep. & Purif. Technol., 86, 2012, p. 70. 8.KISS, A.A., IGNAT, M. R., Applied Energy, 99, 2012, p. 146 9. KISS, A. A., IGNAT, R., FLORES LANDAETA, L., DE HAAN, A.B., Chemical Engineering and Processing: Process Intensification, 67, 2013, p. 39 10.KISS, A. A., IGNAT, M. R., Sep. & Purif. Technol., 98, 2012, p. 290 11.BAKER, R. W., Ind. Eng. Chem. Res., 41, 2002, p. 1393 12. LI, J., ZONG, J., HUANG, W.,XU R., ZANG, Q., SHAO, H., GU, X., Ind. Eng. Chem. Res., 53, 2014, p. 3662 13.HARA, N., YOSHIMUNE, M., NEGISHI, H., HARAYA, K., HARA, S., YAMAGUCHI T., J. Membr. Sci., 450, 2014, p. 215 14.WOINAROSCHY, A., ISOPESCU, R., Ind. Eng. Chem. Res., 49(19), 2010, p. 9195 15.LI, K.H., OLSON, D.H., SEIDEL, J., EMGE, T.J., GONG, H.W., ZENG, H.P., LI, J., J. Am. Chem. Soc. 131, no. 30, 2009, p. 10368. 16.VOß, H., DIEFENBACHER, A., SCHUCH, G., RICHTER, H., VOIGT, I., NOACK, M., CAROD, J., J. Membr. Sci., 329, 2009, p. 11 17.MORI, N., TOMITA, T., Micropor. Mesopor. Mat., 112, no. 1-3, 2008, p. 88 18. XU, L., RUNGTA, M., BRAYDEN, M.K., MARTINEZ, M.V., STEARS B.A., BARBAY G.A., KOROS W.J., J. Membr. Sci. 423 424, 2012, p. 314 19. PAN, Y., PI, T., LESTARI, G., LAI, Z., J. Membr. Sci. 390-391, 2012, p. 93 20.AMARO, C., REQUIÃO, R., EMBIRUÇU, M., International Conference on Modeling, Simulation and Control, 10, 2011, IACSIT Press, Singapore 21.STRÃTULÃ, C., Fracionarea. Principii i metode de calcul, Ed. Tehnicã, Bucuresti 1986 22.NEAGU, M., CURSARU, D. L., Rev. Chim. (Bucharest), 64, no. 8, 2013, p. 880 23.ROMERO, K, Hydrocarbon Processing, April 2012, p. 1 24.DIMA, R., PLE U, V., GÎJIU, C.L., Ingineria separãrilor cu membrane, Ed. Bren, Bucuresti, 1999 25.CHO, Y., KIM, B., KIM, D., HAN, M., LEE, M., J. Proc. Control, 19,2009, p. 932 26.CHILUKURI, P., RADEMAKERS, K., NYMEIJER, van der HAM, L., van den BERG, Ind. Eng. Chem. Res., 46, 2007, p. 8701 Manuscript received: 6.03.2014 264 http://www.revistadechimie.ro REV. CHIM. (Bucharest) 66 No. 2 2015