Extractive distillation for CO2–ethane azeotrope separation

Extractive distillation for CO2–ethane azeotrope separation

Chemical Engineering and Processing 52 (2012) 155–161 Contents lists available at SciVerse ScienceDirect Chemical Engineering and Processing: Proces...

1MB Sizes 0 Downloads 170 Views

Chemical Engineering and Processing 52 (2012) 155–161

Contents lists available at SciVerse ScienceDirect

Chemical Engineering and Processing: Process Intensification journal homepage: www.elsevier.com/locate/cep

Extractive distillation for CO2 –ethane azeotrope separation Fonny Lastari a,∗ , Vishnu Pareek a , Mark Trebble a , Moses O. Tade a , Daniel Chinn b , Nancy C. Tsai c , Kaman I. Chan c a b c

Department of Chemical Engineering, Curtin University of Technology, GPO Box U1987, Perth, Western Australia 6845, Australia Chevron Energy Technology Company, Richmond, CA 94802, USA Chevron Energy Technology Company, Houston, TX 77002, USA

a r t i c l e

i n f o

Article history: Received 7 May 2010 Received in revised form 17 August 2011 Accepted 3 October 2011 Available online 4 November 2011 Keywords: Extractive distillation Simulation Optimization HYSYS CO2 –ethane azeotrope

a b s t r a c t The separation of the CO2 –ethane azeotrope using the hydrocarbon solvents through an extractive distillation process was simulated with the HYSYS 2004.2 software platform. The objective was to examine the optimum solvent amount and composition as well as the optimum feed and solvent location in terms of the overall energy demand. The rigorous simulation results showed that the ratio of the optimum solvent to the minimum solvent amount was in the range of 1.053–1.064 for C4 and C5 solvents in treating the equimolar CO2 –ethane mixture. It was also observed that C4 and C5 solvents were more efficient than C3 solvent due to the lower solvent amount and associated energy demand. The feed and solvent inlet stages have significant effect on the overall energy demand of the column. The best solvent inlet stage is generally near the top of the column, however, an increased solvent loss was observed. © 2011 Elsevier B.V. All rights reserved.

1. Introduction The removal of acid gases (CO2 and H2 S) from natural gas is an essential process not only for increasing the heating value of the liquefied natural gas (LNG) but also to prevent the solidification during the cryogenic processing. There are several methods for acid gas removal, such as chemical absorption with amines, physical absorption, membrane permeation process, and low temperature distillation processes [1]. Of these alternatives, absorption using amine is the most widely used process. However, for high CO2 feedstocks, this process requires a significant amount of solvent and also generates CO2 as a low pressure gaseous product, which must then be either liquefied or compressed for possible geological sequestration. Similarly, the membrane process produces a low pressure CO2 product and also suffers from high pressure drop. The low temperature distillation process, which separates CO2 from hydrocarbons in a series of distillation columns, has the potential to overcome some of the above drawbacks. This process simultaneously produces high pressure CO2 , generates differentiated hydrocarbon products and partly liquefies the LNG [2–4]. There are two significant technical challenges in applying this process: (i) the formation of solid CO2 during the separation of methane in the demethanizer column and (ii) the existence of an azeotrope

∗ Corresponding author. Tel.: +61 8 9266 4687; fax: +61 8 9266 2681. E-mail address: [email protected] (F. Lastari). 0255-2701/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.cep.2011.10.001

in the CO2 recovery column. However, these challenges may be overcome by recycling some of the natural gas liquid (NGL) product, which is produced in the solvent recovery column, as a solvent. The thermodynamics of the process, as well as various flowsheet alternatives, have been investigated in the past and it has been concluded that it is more efficient for processing feedstock with high CO2 and those produced during the enhanced oil recovery (EOR) processes. Several studies for CO2 removal using this method in real plant situations have also been reported in the literature [5,6]. The NGL solvents used in this process typically contain butanes and heavier hydrocarbons, and it acts as a solvent for taking CO2 down the column thereby decreasing the CO2 freezing temperature thus preventing formation of solid CO2 . It also helps in breaking the CO2 –ethane azeotrope in the extractive distillation process. This paper focused only on the operating parameters selection in the extractive distillation column since the separation in the demethanizer is tightly constrained by the CO2 freezing temperature while the extractive distillation process provides more flexibility provided that it is operated above the minimum solvent condition. Fig. 1 is a segment of the process flowsheet showing CO2 and solvent recovery columns. The CO2 recovery column generates pure CO2 in the distillate while ethane and the other heavier hydrocarbons are generated in the bottom. The solvent recovery column provides solvent recycle as well as produces ethane in the distillate stream. A typical feed to the CO2 recovery column is saturated


F. Lastari et al. / Chemical Engineering and Processing 52 (2012) 155–161


y CO2 (mole-fraction)

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9


x CO2 (mole-fraction) Fig. 1. Schematic of CO2 –ethane extractive distillation sequence.


liquid from the bottom of the demethanizer column and contains a CO2 fraction that depends on the level of separation in the demethanizer. The feed enters the CO2 recovery column at an optimum feed inlet stage with solvent entering at a stage near the top of the column. In this study, a majority of simulations were conducted with an equimolar feed of ethane and CO2 ; however to study the effect of feed composition, a range of CO2 –ethane feed compositions were also simulated. Using above simulations, a sensitivity analysis on other operating parameters such as the solvent amount, feed inlet stage, solvent inlet stage, and solvent composition was also conducted. Although butanes and heavier hydrocarbons have been utilized in the past as solvents for this process [2–4], the effect of specific composition of solvents has not been reported. Therefore, in this research we have systematically used both mixed and pure components as solvents. nC5 was first examined since it has a sufficient boiling point difference from the azeotropic constituents as well as phase equilibrium data for CO2 –ethane separation using nC5 solvent are readily available in literature [7]. 2. Separation principle The most important step in an extractive distillation process for the separation of azeotropic or close-boiling components is the addition of an extractive agent that alters the relative volatility of the azeotropic components. The relative volatility (˛) for a high pressure system is calculated as follows [8]: ˛i/j =

˚L /˚V Ki y /x i = i i = iL Kj yj /xj ˚j /˚V j

The degree of alteration depends on the solvent composition and molecular interactions between the key components and the solvent [9]. Fig. 2 shows the altered relative volatility of CO2 –ethane azeotrope as a function of the amount of nC5 solvent added to the azeotrope mixture.





Fig. 2. Phase diagram of CO2 –ethane mixture as a function of nC5 additive.

to a heuristic for conducting simulations on extractive distillation columns which is as follows: 1. Set the column pressure to achieve a specified condensing temperature in the overhead. 2. Choose a typical number of stages for the separation. 3. Specify the solvent amount. In order to establish a converged column, it is suggested to start with a small amount of solvent (10–20% of molar feed flow). 4. Locate the solvent inlet stage near the top of the column, regulated by the maximum solvent limit in the overhead. 5. Place the feed inlet stage in between the solvent inlet stage and the middle section of the column. 6. Choose simple column specifications: CO2 impurity at the bottom and the reflux ratio (for an initial simulation, a reflux ratio between 2 and 4 is typical). 7. Increase the solvent amount gradually until the CO2 overhead specification is achieved. Having the column successfully meeting the desired separation, the reflux ratio column specification is then substituted with the overhead purity specification. 8. Once the first set of convergence with specified component fractions in the overhead and the bottom streams is achieved, the optimum condition for the extractive distillation column can be identified by varying the solvent amount. 4. Results and discussion A fixed number of ideal stages for both columns were used to observe the effect of operating parameters on the reflux ratio and the energy demands. CO2 specifications at the overhead and bottom of the CO2 recovery column were 95 mole% and 40 ppm CO2 , respectively; and the specifications at the overhead and bottom of the solvent recovery column were 1 mole% solvent and 0.1 mole% ethane, respectively.

3. Simulation setup 4.1. Effect of solvent amount The flowsheet in Fig. 1 was simulated using HYSYS 2004.2 (commercial software for process simulations) with Peng–Robinson property package, which is widely used for non-ionic hydrocarbons. The operating parameters for simulations were optimized for minimum energy targets. The design of an extractive distillation column for CO2 –ethane azeotrope separation often encounters difficult convergence problems in HYSYS. However, an analysis of rigorous simulations led

As shown in Fig. 2, a minimum quantity of solvent is required to keep the CO2 –ethane equilibrium line (solvent-free basis) away from the y = x line (i.e., the azeotropic point). For fixed azeotropic feed and specified product purity, the minimum amount of solvent is dictated by the vapour liquid equilibrium in the column, which is a function of the operating parameters, such as pressure and temperature, the feed inlet stage and the solvent inlet stage. For fixed

F. Lastari et al. / Chemical Engineering and Processing 52 (2012) 155–161


Fig. 3. CO2 purity in the overhead as a function of the solvent amount (N = 50, Nf = 29, Ns = 3, condenser pressure = 2415 kPa and condenser temperature = −14 ◦ C).

feed and solvent inlet stages, Fig. 3 shows that below a minimum solvent amount (solvent to feed rate ratio, S/Fmin = 0.57), the desired purity of 95% CO2 in the overhead was not achievable regardless of the reflux ratio applied. When the minimum solvent amount was used, the required CO2 purity was achieved at a reflux ratio of about 3.3. As the reflux ratio was increased above this value, the CO2 purity declined. This trend was also observed for all other solvent amounts. This peculiar behaviour is typical of extractive distillation processes as excessive reflux ratio effectively dilutes the solvent thus worsening the separation [10]. When the S/F ratio was greater than the minimum, S/Fmin , the optimum reflux ratio decreased moderately as shown in Fig. 3. These two variables, i.e. the solvent quantity and the reflux ratio, both have a direct consequence on the energy requirement. Therefore, to decide the most economical solvent amount, an analysis on the energy requirements for the process was conducted. Fig. 4 depicts the effect of solvent amount on the total energy demands in both extractive distillation and solvent recovery (SR) columns shown in Fig. 1. In the extractive/azeo column, the condenser duty (EEAZCOND) continuously decreased with the solvent amount; the reboiler duty (EEAZREB), on the other hand, first declined due to the lower reflux, but then rose again due to the increase in the solvent flow rate (Fig. 4a). However, in the solvent recovery column, both the condenser and reboiler duties increased with the solvent amounts (Fig. 4b). Fig. 4c plots a cumulative effect of solvent flow rate on the total condenser and reboiler duties in the two columns and the required reflux ratio in the azeo column. It is clear from these plots that the solvent amount should strike an economic balance between the condenser and reboiler duties for both extractive and solvent recovery columns. From Fig. 4c, the optimum solvent amount for a minimum total energy demand is 11,200 kg mole/h. The ratio of the optimum to the minimum solvent in this case was 1.053. Fig. 5 shows the typical composition profile of an extractive distillation column. The liquid profile in the rectifying and stripping sections start at the distillate (pure CO2 ) end and bottom (C2–C5) side, respectively. The vapour profile moves along the CO2 –C2 side (low solvent content). It was observed that, at the minimum solvent amount, the nC5 liquid composition profile was practically constant along the column; however, with excess solvent, the nC5 liquid fraction varied considerably from the top to the bottom of the column. Thus, the nC5 liquid profile can be used to indicate if the column is operating with an efficient amount of solvent. Furthermore, as more solvent was employed, the nC5 fraction in the vapour phase was higher and thus more solvent was lost in the overhead product.

Fig. 4. (a) Effect of the solvent amount on the condenser and reboiler duties in the extractive column. (b) Effect of the solvent amount on the condenser and reboiler duties in the solvent recovery column. (c) Effect of the solvent amount on the reflux ratio and total energy demand.


F. Lastari et al. / Chemical Engineering and Processing 52 (2012) 155–161

Fig. 5. Composition profiles as a function of solvent amount (䊉 = liquid profile;  = vapour profile with minimum solvent 10,636 kg mole/h; profile with solvent 12,475 kg mole/h;  = liquid profile;  = vapour profile with excess solvent 24,950 kg mole/h).

All of the simulations performed displayed similar composition profiles to those shown in Fig. 5. The rectifying section started at the pure CO2 end and the stripping section started at the bottom (C2–C5) side, depending on the mass balance. The extractive section acted as a bridge connecting the rectifying section to the stripping section by passing through the triangular space. The extractive section could be near the nC5 node when a high solvent flow and low reflux were utilized; however, it could be located near the CO2 –C2 side when minimum solvent amount or a high reflux was applied. 4.2. Effect of feed inlet stage

= vapour

was 0.019, and the CO2 and nC5 liquid fractions had different values for different minimum solvent amounts. For a minimum energy requirement, Fig. 6 also indicates that there is an optimum feed inlet stage for both the solvent amount and the reflux ratio. In this study, the optimum feed inlet stage was found to be stage 26. 4.3. Effect of solvent inlet stage As shown in Fig. 7, the solvent inlet stage has a direct impact on the reflux ratio and the solvent impurity in the overhead product. As the solvent inlet stage was lowered, simulation showed that a higher reflux ratio was required. This was due to a decrease in the nC5 solvent in the liquid phase above the solvent inlet stage. Although moving the solvent inlet stage upwards decreased the reflex ratio but at the same time increased the amount of nC5 solvent going to the overhead. Therefore, the solvent stage should be optimally determined not only to decrease the reflux ratio but also to prevent excessive amount of solvent going to the overhead. In this study, the maximum nC5 fraction was specified as 100 ppm. With this limit, Fig. 7 shows that the optimum solvent inlet location was stage 3 from the top. 4.4. Effect of CO2 composition













3.00 14









Feed Inlet Stage Minimum Solvent Flow

Reflux Ratio

Fig. 6. Reflux ratio and minimum solvent amount as a function of feed inlet stage.

Solvent Inlet Stage 1.00E+00 1.00E-01 0 1.00E-02 1.00E-03 1.00E-04 1.00E-05 1.00E-06 1.00E-07 1.00E-08 1.00E-09








Mole fraction nC5 in overhead Fig. 7. Effect of solvent inlet stage.


2.90 9 2.85 2.80 2.75 2.70 2.65 2.60 2.55 2.50 2.45

Reflux Ratio

Reflux ratio


Mole fraction nC5 in overhead

Fig. 8 shows the amount of solvent and reflux ratio demands for various CO2 feeds for an extractive distillation column with 50 theoretical stages and the same product specifications (i.e. 95% CO2

Reflux Ratio

Minimum Solvent Flow (kgmole/hr)

The location of the feed inlet stage strongly affects the required amount of solvent and the reflux ratio. Fig. 6 shows the minimum solvent amount and the related optimum reflux ratio as a function of the feed inlet stage. For a column with 50 total theoretical stages, the lowest minimum solvent demand was observed when the feed inlet was at stage 29. When the feed inlet stage was moved, the reflux ratio was also altered to maintain the compositional product specifications. As a result of the altered reflux ratio, the solvent amount, which is governed by the vapour liquid equilibrium at the solvent inlet stage, was also changed. However, at the solvent inlet stage, it was observed that the C2, CO2 , and nC5 vapour fractions were 0.021, 0.971, and 0.008, respectively, and the C2 liquid fraction

= liquid profile;

60000 50000 40000 30000 20000 10000 0 0








Reflux Ratio 20% CO2

35% CO2

50% CO2

65% CO2

Mole fraction of solvent to break CO2-C2 azeotrope

Minimum Solvent Flow (kgmole/h)

F. Lastari et al. / Chemical Engineering and Processing 52 (2012) 155–161




0.22 0.2








0.16 0.14 0.12 C3



Fig. 9. Mole fraction of the solvent required to break the CO2 –ethane azeotrope. Fig. 8. Effect of CO2 content in the feed on the reflux ratio and minimum solvent amount.

purity in the overhead and 40 ppm mole CO2 in the bottom product). At low reflux ratios, the solvent amount decreased with the CO2 fraction in the feed, while an opposite behaviour is observed for higher reflux ratios. This particular behaviour explains that the solvent quantity is a function of the amount of ethane in the column. At lower reflux ratios, a higher CO2 content in the feed means less C2 and therefore, a lower solvent quantity is required. At higher reflux ratios, although the C2 fraction in the feed decreases with CO2 fraction, the liquid flow in the column still contains higher amount of C2 because of the increased reflux flow and thus needing a higher solvent amount. 4.5. Single component solvents Using ethane–CO2 pseudo-binary diagrams (on solvent free basis, see Fig. 2 for example), minimum solvent demand for each hydrocarbon solvent was calculated which is shown in Fig. 9. It is clear from this figure that the solvent demand is lower with heavier hydrocarbon as the solvent. Fig. 10 displays the minimum solvent amount and the reflux ratio with single component solvents. For the minimum amount of solvent, the results are similar to those in Fig. 9 with solvent demand being higher for lower hydrocarbons. This increased requirement of solvent for lighter hydrocarbons poses further problems due to their higher relative volatility, which increases their composition in the overhead product. Therefore, in order to ensure the required product purity, a significantly higher reflux ratio is needed with C3 as the solvent. These observations lead to the conclusion that C3 is not a desirable solvent for CO2 –ethane separation.

Table 1 Ratio of the optimum solvent to the minimum solvent for each single component solvents. Component

Ratio of optimum solvent to minimum solvent

Energy requirement with optimum solvent amount (kJ/h)

C3 iC4 nC4 iC5 nC5

1.052 1.064 1.063 1.054 1.053

1.394 × 109 1.077 × 109 1.055 × 109 1.130 × 109 1.147 × 109

Further simulations were performed for each of the hydrocarbon solvents to determine the location of optimum feed and solvent inlet stages using the same procedure as for the nC5 solvent (Figs. 6 and 7). It is clear from Fig. 10 that for heavier solvents, optimal solvent stage is stage 3 but for lighter solvents (C3), optimal solvent inlet is stage 6. Even after this further optimization, the amount needed for C3 solvent was significantly higher than that for C5 solvent. Table 1 shows energy requirements as a function of solvent type and amount (reported as the ratio of solvent amount to minimum solvent). From Table 1, it is evident that the ratios of the optimum solvent to the minimum solvent amounts are similar for all of the hydrocarbon solvents; however, the energy demand is highest with C3 solvent due to the higher reflux ratio requirements in the both extractive and the solvent recovery columns. 4.6. Effect of solvent composition Following the single component solvent study above, the effect of solvent composition was investigated by varying the solvent

Fig. 10. Reflux ratio and minimum solvent for different single component solvents.


F. Lastari et al. / Chemical Engineering and Processing 52 (2012) 155–161

Fig. 11. Minimum solvent amount required for different solvent compositions.

Fig. 12. Effect of the solvent composition on the reflux ratio.

composition under the same column configuration (i.e., fixed number of stages, feed inlet stage, and solvent inlet stage). The effect of this parameter on the minimum solvent amount, reflux ratio, and the total energy requirement is shown in Figs. 11–13. A “state” in

these figures refers to a solvent mixture with a specified composition. As shown in Fig. 11, the minimum solvent quantity is lower for the solvents with smaller C3 fraction. The lowest solvent demand

Fig. 13. Total energy requirement as a function the solvent composition.

F. Lastari et al. / Chemical Engineering and Processing 52 (2012) 155–161 Table 2 CO2 recovery column overhead product composition.

mole% C2 mole% total solvent mole% CO2

State 1

State 2

State 3

State 4

State 5

State 6

0.030 0.020 0.950

0.038 0.012 0.950

0.043 0.007 0.950

0.049 0.001 0.950

0.048 0.002 0.950

0.050 0.00004 0.950


For mixed component solvents, a higher fraction of C3 in the solvent needed higher reflux ratios and a higher minimum solvent amount. Finally, for minimum energy demands, the C4 composite solvent was found to be the most optimal solvent.

Acknowledgement was found with the C5 mixture solvent. A similar trend for the reflux ratio was also obtained as depicted in Fig. 12. Fig. 13 shows the effect of the solvent composition on the total energy requirement in the both extractive and the solvent recovery columns. It was observed that the lowest energy demand was found with the C4 solvent. The higher energy demand with the C5 solvent could be attributed to corresponding higher reboiler duties in the columns. It is also important to observe the loss of valuable hydrocarbon in the CO2 overhead product. Table 2 shows that, with a specified CO2 level in the overhead, the solvent loss is lower when heavier hydrocarbon solvents are used. In this study, out of several solvent compositions, the C4 composite solvent was the best option in terms of the energy demand. 5. Conclusions In this study, extensive process simulations were performed on an extractive distillation circuit using HYSYS. Simulation results showed that for an extractive distillation column, there exists a minimum solvent amount and an optimum solvent amount. The minimum solvent amount indicates the quantity required to achieve the product specifications, whereas the optimum solvent amount is associated with a minimum energy demand. For the pure nC5 solvent, the ratio between the optimum and minimum solvent amounts was 1.053. The inlet feed and solvent stages also had a significant effect on the overall energy demand for the column. The best solvent inlet stage is generally near the top of the column, but may not be the top tray as it will result in increased loss of the solvent in the overhead CO2 product. Furthermore, the minimum solvent amount was also a function of CO2 composition in the feed and the reflux ratio. The solvent demand was also related to the amount of C2 that needs to be extracted by the solvent. From the single component solvent study, it was observed that propane is not a suitable solvent for CO2 –ethane separation while C4 and C5 yield similar results in terms of the amount of solvent required and the associated energy demand. The ratio of the optimum solvent amount to the minimum amount for C4 and C5 solvents were in the range of 1.053–1.064.

The financial support for this research from the Chevron Energy Technology Company through the WA:ERA (Western Australian Energy Research Alliance) Gas Research Program is gratefully acknowledged. Appendix A. Nomenclature

K y x ˚L ˚V i j N Nf Ns

distribution coefficient vapour phase mole fraction liquid phase mole fraction activity coefficient in liquid phase fugacity coefficient in vapour phase component i component j number of equilibrium stages feed inlet stage solvent inlet stage

References [1] A. Kohl, R. Nielsen, Gas Purification, 5th edition, Gulf Publishing Company, Houston, 1997. [2] A.S. Holmes, J.M. Ryan, B.C. Price, R.E. Styring, Process improves acid gas separation, Hydrocarbon Process. (May) (1982) 131–136. [3] J.M. Ryan, F.W. Schaffert, CO2 recovery by the Ryan/Holmes process, Chem. Eng. Prog. (October) (1984) 53–56. [4] F.W. Schaffert, J.M. Ryan, Ryan/Holmes technology lands EOR projects, Oil Gas J. (January) (1985) 133–136. [5] B.C. Price, F.L. Gregg, CO2 /EOR: from source to resource, Oil Gas J. (August) (1983) 116–122. [6] N.V. Wood, J.V. O’Brien, F.W. Schaffert, Seminole CO2 flood gas plant has successful early operations, Oil Gas J. (November) (1986) 50–54. [7] J.H. Hong, R. Kobayashi, To break an azeotrope. The use of n-pentane to break the CO2 –ethane azeotrope, for CO2 EOR gas processing, Ind. Eng. Chem. Process Des. Dev. 25 (1986) 736–741. [8] R.C. Reid, J.M. Prausnitz, B.E. Poling, The Properties of Gases and Liquids, 4th edition, McGraw Hill Companies, Inc., Mexico, 1987. [9] J.D. Seader, J.J. Siirola, S.D. Barnicki, Perry’s Chemical Engineering Handbook, 7th edition, McGraw Hill Companies Inc., USA, 1997, pp. 13-75–13-81. [10] L. Laroche, N. Bekiaris, H.W. Andersen, M. Morari, The curious behaviour of homogeneous azeotropic distillation – implications for entrainer selection, AIChE J. 38 (9) (1992) 1309–1328.