- Email: [email protected]
Rigorous simulation of energy integrated and thermally coupled distillation schemes for ternary mixture
Applied Thermal Engineering 21 (2001) 1299±1317
www.elsevier.com/locate/apthermeng
Rigorous simulation of energy integrated and thermally coupled distillation schemes for ternary mixture Mansour Emtir, Endre Rev *, Zsolt Fonyo Department of Chemical Engineering, Budapest University of Technology and Economics, H-1521 Budapest, Hungary Received 28 November 2000; accepted 22 January 2001
Abstract Sharp (99% purity) separation of a ternary mixture, characterised by near uniformly distributed volatility, by direct separation sequence without, with forward, and with backward energy integration, by indirect separation sequence without, with forward, and with backward energy integration, by sloppy separation sequence with forward and with backward energy integration, and by thermally coupled sloppy separation sequence (Petlyuk system) is rigorously modelled and optimised. Three feed compositions, namely (Case 1) equimolar A/C ratio with 10% B, (Case 2) equimolar feed, and (Case 3) equimolar A/C ratio with 80% B are compared. Comparison is based on total annual costs (TACs) using European and American price systems. The savings in TAC of Petlyuk are uniformly about 28±33% in all the three cases, while the savings of the energy integrated systems increases together with increasing ratio of B in the feed. In Case 1, Petlyuk system is the winner, with 33% savings, but a conventional energy-integrated system is handicapped by just a very few percent. In Cases 2 and 3, Petlyuk system is not amongst the best structures. In Case 2, either the conventional energy-integrated systems or the energy-integrated sloppy structures win with 35±41%; while Case 3, the energy-integrated sloppy structures are the best with about 51% savings. Ó 2001 Elsevier Science Ltd. All rights reserved.
1. Introduction Distillation is the most widely used separation technique in the petrochemical and chemical process industries for the separation of ¯uid mixtures despite its high energy requirement. Signi®cant energy savings can be reached by the use of complex distillation arrangements such as the side-stripper, the side-recti®er, the thermal (internal) column coupling (also known as Petlyuk
*
Corresponding author. Fax: +36-1-463-3197. E-mail address: [email protected] (E. Rev).
1359-4311/01/$ - see front matter Ó 2001 Elsevier Science Ltd. All rights reserved. PII: S 1 3 5 9 - 4 3 1 1 ( 0 1 ) 0 0 0 1 7 - 5
1300
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
Nomenclature A A B C D D DQF DQB Eo ESI hW hl H I IQF IQB Lm
heat transfer area light component middle component heavy component column diameter (m) conventional direct sequence direct sequence with forward energy integration direct sequence with backward energy integration overall plate eciency ease of separation index, ESI aAB =aBC weir height (m) liquid height above valve opening (m) column height (m) conventional indirect sequence indirect sequence with forward energy integration indirect sequence with backward energy integration molar liquid ¯ow rate (kmol/h)
M&S N P Q S SQD SQB SQF SP T TAC U Vm x
Marshall and Swift index number of plates pressure heat duty (kJ/h) sloppy sequence without integration or coupling (Pre¯ash system) sloppy sequence with double energy integration sloppy sequence with backward energy integration sloppy sequence with forward energy integration thermally coupled sloppy sequence (Petlyuk system) temperature total annual costs overall heat transfer coecient molar vapor ¯ow rate (kmol/h) mole fraction
Greeks a Relative volatility k Average latent heat Average viscosity of feed (cP) lavg
system), the (external) energy integration (also known as energy integrated distillation system) and the heat pumping techniques. Theoretical studies, [1,2,6,8,9] have shown that the column coupling con®gurations are capable of achieving typically 30% of energy savings compared to a conventional sequence. In addition, the coupling con®guration can also be achieved with the so called dividing wall column, by placing a vertical wall in the middle of the column separating the feed from the side draw, e.g. [3,10]. With this arrangement, reduction in capital cost can also be expected through the elimination of one of the column shells (but not the column internal). Despite the above advantages, industry has been reluctant to use the Petlyuk system and dividing wall columns and this is usually attributed to the lack of established design procedures and the fear of control problems. This paper compares the energy integrated and thermally coupled con®gurations based on rigorous simulation and optimisation for total annual costs (TACs). In this article the following abbreviations are used for distillation structures, including not only those rigorously optimised but also those studied, by short-cut methodology, in our earlier publication [7]:
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
D DQF DQB I IQF IQB S SQF SQB SQD SP
1301
Direct separation sequence without energy integration (Fig. 1). Direct separation sequence with forward energy integration (Fig. 2). Direct separation sequence with backward energy integration (Fig. 3). Indirect separation sequence without energy integration (Fig. 4). Indirect separation sequence with forward energy integration (Fig. 5). Indirect separation sequence with backward energy integration (Fig. 6). Sloppy separation sequence without any energy integration or thermal coupling (Fig. 7). Sloppy separation sequence with forward energy integration (Fig. 8). Sloppy separation sequence with backward energy integration (Fig. 9). Sloppy separation sequence with double (both forward and backward) energy integration (Fig. 10). Thermally coupled sloppy separation sequence (Petlyuk system) (Fig. 11).
The sloppy separation with or without a single energy integration (S, SQF, and SQB) can be realised by three columns or with a system of a pre¯ash column and a side-stream column obtained by lumping the second and third columns. The double integrated sloppy structure SQD is
Fig. 1. Direct separation sequence without energy integration (D).
Fig. 2. Direct separation sequence with forward energy integration (DQF).
1302
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
Fig. 3. Direct separation sequence with backward energy integration (DQB).
Fig. 4. Indirect separation sequence without energy integration (I).
Fig. 5. Indirect separation sequence with forward energy integration (IQF).
realized by a three-column system. Lumping the two end columns is a precondition for realizing the Petlyuk structure SP.
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
1303
Fig. 6. Indirect separation sequence with backward energy integration (IQB).
Fig. 7. Sloppy separation sequence without any energy integration or thermal coupling (S).
According to our previously performed short-cut analysis [7], in the case of sharp separation the sloppy separation path with forward, backward, or double energy integration (SQF, SQB, or SQD) is almost always capable of achieving as much energy savings as that of with the thermal coupling (Petlyuk system or dividing wall column, SP). Moreover, the conventional direct sequence and indirect sequence structures with energy integration (DQF, DQB, IQF, IQB) are, in some cases, also capable of achieving larger energy savings than that of the corresponding Petlyuk or dividing wall column con®gurations (SP).
1304
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
Fig. 8. Sloppy separation sequence with forward energy integration (SQF).
Fig. 9. Sloppy separation sequence with backward energy integration (SQB).
The main corollaries of the short-cut analysis are the following: From energy point of view the integrated and coupled structures are always better than the non-integrated ones. From energy point of view the integrated or thermally coupled structures almost always win. There is just a very small domain for DQ to win at very high ratio of A in the feed. All the integrated and coupled sloppy structures are equivalent. Considering exergy loss, the integrated and coupled structures share the Gibbs composition triangle. There is a signi®cant area where SP wins near the AC edge. According to the short-cut analysis, the forward or backward energy-integrated, as well as the double energy-integrated sloppy structures (SQF, SQB, and SQD) are equivalent, in energy term,
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
1305
Fig. 10. Sloppy separation sequence with double (both forward and backward) energy integration (SQD).
Fig. 11. Thermally coupled sloppy separation sequence (Petlyuk system) (SP).
to the thermally coupled sloppy structures (PS, i.e. Petlyuk or dividing wall column con®guration); and the energy-integrated structures win, in energy term, almost everywhere of the studied conditions. These include a wide range of relative volatility ratios. The Petlyuk structure has the greatest chance to win over the energy-integrated schemes at balanced relative volatility ratio. Even in that case, the integrated structures win over the Petlyuk system in the greatest part of the feed composition triangle; but the Petlyuk system yet proves to being a winner in a signi®cant area of feed compositions.
1306
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
2. Rigorous case studies Continuous recti®cation of a ternary mixture of ethanol, n-propanol, and n-butanol at 99% product purity speci®cations and three dierent feed compositions are simulated and optimized in this study. The feed compositions, candidates for the SP con®guration to win, are selected according to the results of the short-cut analysis. The main data of the mixtures to be separated are listed in Tables 1±3.
Table 1 Feed and product speci®cations for Case 1 Streams
Feed
Ethanol
n-Propanol
n-Butanol
kmol/h
Fraction
kmol/h
Fraction
kmol/h
Fraction
kmol/h
Fraction
Components Ethanol (A) n-Propanol (B) n-Butanol (C)
135 30 135
0.45 0.1 0.45
134.87 1.36 0.00
0.99 0.01 0.00
0.14 27.27 0.14
0.005 0.990 0.005
0.00 1.36 134.87
0.00 0.01 0.99
Total
300
1.000
136.23
1.00
27.55
1.000
136.23
1.00
Table 2 Feed and product speci®cations for Case 2 Streams
Feed
Ethanol
kmol/h
Fraction
Components Ethanol (A) n-Propanol (B) n-Butanol (C)
100 100 100
0.333 0.333 0.333
Total
300
1.000
kmol/h
n-Propanol
n-Butanol
Fraction
kmol/h
Fraction
kmol/h
Fraction
99.50 1.01 0.00
0.99 0.01 0.00
0.49 98.00 0.49
0.005 0.990 0.005
0.00 1.01 99.50
0.00 0.01 0.99
100.51
1.00
98.98
1.000
100.51
1.00
Table 3 Feed and product speci®cations for Case 3 Streams
Feed
Ethanol
n-Propanol
n-Butanol
kmol/h
Fraction
kmol/h
Fraction
kmol/h
Fraction
kmol/h
Fraction
Components Ethanol (A) n-Propanol (B) n-Butanol (C)
30 240 30
0.10 0.80 0.10
28.79 0.29 0.00
0.99 0.01 0.00
1.21 239.42 1.21
0.005 0.990 0.005
0.00 1.01 99.50
0.00 0.01 0.99
Total
300
1.000
29.08
1.00
241.84
1.000
29.08
1.00
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
1307
Table 4 Relative volatilities and ease of separation indices aAB aAC aBC ESI
Case 1
Case 2
Case 3
2.02 4.67 2.31 0.87
2.07 4.72 2.29 0.90
2.15 4.79 2.23 0.96
Fig. 12. Composition triangle with the location of the feed compositions.
The relative volatilities are approximately balanced in this material system. The mean relative volatilities and estimated ease of separation are shown in Table 4. The feed points are located near the centre of the AC edge in Case 1, in the centre of the Gibbs composition triangle Case 2, and near the node of pure B in Case 3 (see Fig. 12). In either case, 99% product purity and equal distribution of the impurities is speci®ed in all the products. The D, DQF, DQB, I, IQF, IQB, S, SQF, SQB, and SP (Petlyuk) structures are simulated, optimized, and then compared. The neither integrated nor thermally coupled sloppy system (the so-called pre¯ash system) with three columns (S) and the double integrated sloppy structure (SQD) are not studied rigorously; the ®rst one because of its evident inferiority, the second one because of its expected excess complexity in realization. The costs of the dividing wall column (a variant of SP) is, on the other hand, not determined; the Petlyuk system is studied, instead. The two conventional systems D and I are studied just for obtaining a base case to determine the savings. HYSYS simulation package is applied for rigorous modelling in all studied systems with NRTL thermodynamic property model using the data sets built in the simulator, atmospheric feed and products at saturated liquid state. Two radically dierent cost structures are taken into account. One of them is a high utility price structure corresponding to prices in Europe; the other one is a low utility price structure that corresponds to prices in the USA. Utility cost data are collected in Table 5. The eect of the high and low utility costs is demonstrated in Cases 1 and 2; therefore, Case 3 is studied with the European cost structure only. Valve trays (of Glitsch type) are considered as column internal. Installation cost formulas [5] and data are collected in Appendix A.
1308
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
Table 5 Utilities cost data Utility LP-steam MP-steam Cooling water Electricity a b
High utility pricesa
Low utility pricesb
Temperature level (°C)
Price
Temperature level (°C)
Price
160 184 30±45 ±
17.7 $/ton 21.8 $/ton 0.027 $/ton 0.1 $/kW h
160 184 30±45 ±
6.62 $/ton 7.31 $/ton 0.0067 $/ton 0.06 $/kW h
Based on European prices. Based on USA prices.
Results of Case 1 are collected in Tables 6 and 7. Here the feed is situated near the AC edge. The presence of component B in the feed is just 10%; the presence of components A and B are equimolar to each other. The two tables correspond to the two subcases of American and European utility prices. The winners and their rate of winning are collected in Table 8. SP wins, in TAC savings compared to the base case D, with 4% and 3% above DQB, and with 7% and 6% above IQF. The Petlyuk system is better than the others in both of its lower operating costs and capital costs are. According to the short-cut analysis, at this feed composition the Petlyuk structure (SP) ought to win over one of the integrated structures with approximately 40%. According to Tables 6 and 7, however, in case of the lower utility costs (American price structure) the handicap of DQB behind SP is just 5.7% in TAC, 8% in capital costs, and 4.8% in energy costs. In case of the higher utility costs (European price structure), the handicap of DQB behind SP is again no more than 4.85% in TAC, 7.8% in capital costs, and 5.55% in energy costs. Results of Case 2 are collected in Tables 9 and 10. Here the feed is equimolar in the components. The winners and their rate of winning are collected in Table 11. According to the short-cut analysis, at this feed composition the Petlyuk structure (SP) ought to win over one of the integrated structures with approximately 20%. On the contrary, SP does not win in any of the two rigorous subcases. We also investigated these cases with sieve trays, and got the same qualitative results (not listed here). Instead of SP, the integrated sloppy structures SQF and SQB win in the case of American utility prices. However, they are just 1% better above DQB, in TAC savings compared to the base case D. According to Tables 9 and 10, in case of the lower utility costs (American price structure) the handicap of DQB behind SQF and SQB is just 1.8% and 0.8% in TAC, 4.1% and 3.5% in capital costs, and 0.87% and 0.57% in energy costs. (In the latest ®gure, DQB is better in energy costs over SQB.) In case of the higher utility costs (European price structure) the structures change place. Here DQB wins over the sloppy structures. The handicap of SQF and SQB behind DQB is 7.3% and 8.2% in TAC, and 9.3% and 11.5% in energy costs, while they have better capital cost ®gures with 3.2% and 4.5%. The Petlyuk system (SP) is not in the ®rst ®ve places in the case of the American price structure; and it is just at the 4th place, with great handicap, in the case of the higher (European) utility prices.
a
155 1:98E 07 1:93E 07 ± 5:04E 05 1:65E 04 5:21E 05 1:19E 05 6:40E 05 0 0 0
DQF
IQF
DQB
IQB
SP
SQF
SQB
101.33 1.35 7.73 47
130 1:53E 07 1:36E 07 8:04E 06 4:47E 05 1:17E 04 4:58E 05 1:38E 05 5:97E 05 16 12 7
512.50 1.39 1.50 83
101.33 1.36 1.30 53
124 1:37E 07 1:21E 07 1:14E 07 3:49E 05 1:03E 04 3:59E 05 1:45E 05 5:04E 05 22 31 21
249.00 1.42 0.92 71
230.50 1.20 6.36 50
128 1:33E 07 1:19E 07 8:01E 06 3:39E 05 1:01E 04 3:49E 05 1:35E 05 4:84E 05 13 33 24
101.33 1.49 1.26 78
512.00 1.23 1.62 52
123 1:38E 07 1:19E 07 1:22E 07 3:51E 05 1:02E 04 3:61E 05 1:50E 05 5:11E 05 26 31 20
101.33 1.51 0.85 71
101.33 1.54 1.32 126
187 1:27E 07 1:22E 07 ± 3:23E 05 1:04E 04 3:33E 05 1:25E 05 4:58E 05 5 36 28
101.33 1.31 0.64 61
101.33 1.24 0.94 114 166 1:22E 07 1:17E 07 7:15E 06 3:57E 05 9:98E 03 3:67E 05 1:37E 05 5:04E 05 15 30 21
495.00 1.22 0.74 52
458.00 1.28 1.06 105 158 1:27E 07 9:21E 06 9:66E 06 3:70E 05 7:88E 03 3:78E 05 1:43E 05 5:21E 05 20 27 19
101.33 1.31 0.63 53
Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 101.33 1.34 1.10 85
156 2:34E 07 2:29E 07 ± 5:95E 05 1:96E 04 6:15E 05 1:28E 05 7:43E 05 8 18 16
101.33 1.51 0.85 71
Col. 1
101.33 1.21 5.46 75
Col. 2
Col. 1
101.33 1.49 1.26 80
Ia
D
If the top product of column 1 (AB) is taken and fed to column 2 as vapor then TAC ($/yr) is 6.04E 05.
Total actual plates Heating rate (kJ/h) Cooling rate (kJ/h) Main HX duty (kJ/h) Steam cost ($/yr) C.W cost ($/yr) Operat. cost ($/yr) Capital cost ($/yr) TAC ($/yr) Capital savings (%) Op. cost savings (%) TAC savings (%)
Pressure (kPa) Diameter (m) Re¯ux ratio Actual plates
Description
Table 6 Case 1 optimal schemes based on US utility prices
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317 1309
a
IQF
DQB
IQB
SP
SQF
SQB
176 2:32E 07 2:27E 07 ± 1:58E 06 7:87E 04 1:66E 06 1:37E 05 1:79E 06 12 18 17
101.33 1.34 1.09 93
101.33 1.35 7.73 47
143 1:52E 07 1:36E 07 8:04E 06 1:32E 06 4:70E 04 1:37E 06 1:43E 05 1:51E 06 16 3 1
512.50 1.37 1.49 96
101.33 1.36 1.30 53
137 1:36E 07 1:21E 07 1:14E 07 9:24E 05 4:19E 04 9:66E 05 1:50E 05 1:12E 06 22 31 27
249.50 1.41 0.90 84
230.50 1.20 6.36 50
136 1:32E 07 1:18E 07 8:01E 06 9:01E 05 4:09E 04 9:42E 05 1:38E 05 1:08E 06 12 33 30
101.33 1.48 1.24 86
512.00 1.23 1.62 52
135 1:37E 07 1:17E 07 1:22E 07 9:30E 05 4:07E 04 9:70E 05 1:55E 05 1:13E 06 27 31 27
101.33 1.51 0.82 83
101.33 1.53 1.31 128
195 1:26E 07 1:21E 07 ± 8:59E 05 4:20E 04 9:01E 05 1:28E 05 1:03E 06 4 36 33
101.33 1.30 0.63 67
101.33 1.24 0.93 116 172 1:22E 07 1:16E 07 7:09E 06 1:06E 06 4:04E 04 1:10E 06 1:39E 05 1:24E 06 13 22 19
495.00 1.21 0.73 56
457.00 1.28 1.06 105 160 1:27E 07 9:20E 06 9:64E 06 1:10E 06 3:19E 04 1:13E 06 1:43E 05 1:28E 06 17 19 17
101.33 1.31 0.63 55
Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2
DQF
If the top product of column 1 (AB) is taken and fed to column 2 as vapor then TAC ($/yr) is 1:42E 06.
166 1:97E 07 1:92E 07 ± 1:34E 06 6:67E 04 1:41E 06 1:22E 05 1:53E 06 0 0 0
101.33 1.51 0.82 83
Col. 1
101.33 1.20 5.39 86
Col. 2
Col. 1
101.33 1.49 1.26 80
Ia
D
Total actual plates Heating rate (kJ/h) Cooling rate (kJ/h) Main HX duty (kJ/h) Steam cost ($/yr) C.W cost ($/yr) Operat. cost ($/yr) Capital cost ($/yr) TAC ($/yr) Capital savings (%) Op. cost savings (%) TAC savings (%)
Pressure (kPa) Diameter (m) Re¯ux ratio Actual plates
Description
Table 7 Case 1 optimal schemes based on European utility prices
1310 M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
1311
Table 8 Winning structures in Case 1 (10% middle component) USA prices
EU prices
Structure
Savings (%)
Structure
Savings (%)
SP DQB IQF SQF IQB
28 24 21 21 20
SP DQB IQF IQB SQF
33 30 27 27 19
Results of Case 3 are collected in Table 12. Here the feed is situated near the B node. The presence of component B in the feed is 80%; the presence of components A and B are equimolar to each other. Just the subcase of the European utility prices is shown. As the ratio of component B increases in the feed, the energy demand of the separation increases, and the energy costs become dominating. The structures DQF, IQF, and IQB are proved inferior to the others, as the energy costs are dominant; therefore, they are omitted. The conventional indirect structure I is, however included for comparison. The winners and their rate of winning are collected in Table 13. In accordance with the results of the short-cut analysis, this is the place where the integrated sloppy structures (SQB and SQF) win. The Petlyuk system (SP) is again at the fourth place behind the third rank DQB. The handicap of the DQB system behind SQB and SQF is 44% and 38% in TAC, 4.1% and 5.7% in capital costs, and 52% and 43% in energy costs.
3. Conclusions A general rule governing the ranks of the studied structures according to TAC is the increasing heat duty requirement with increasing concentration of component B in the feed. The structures with energy integration are sensitive for the heat duty while the Petlyuk system without energy integration (but thermal coupling) is not. The savings in TAC of Petlyuk are uniformly about 30± 33% in all the three cases. Case 1: Here the Petlyuk system (PS) is the winner with 28% or 33% savings. There is no qualitative ranking dierence considering the two price structures. Second ranked is the direct sequence with backward energy integration (DQB) with 24% or 30% savings that cannot be more because of low heat duty. It is back just with 3%. The forward energy integrated sloppy system (SQF, 19% savings) is just the third or fourth ranked, probably because there is an AB ! C energy integration that involves a pressure shift and use of high pressure steam. Case 2, American prices: Here the Petlyuk system is not amongst the best structures. Direct sequence with backward energy integration (DQB) with 35% savings is at the third place backed just with 1% by both forward energy integrated sloppy system (SQF) and backward energy integrated sloppy system (SQB) by their 36% savings. Case 2, European prices: Here the direct sequence with backward integration (DQB) is the winner with 41% savings. Both the second ranked forward integrated sloppy system (SQF, 36%
a
Ia DQF
IQF
DQB
IQB
SP
SQF
SQB
101.33 1.50 1.82 77
156 2:43E 07 2:39E 07 ± 6:17E 05 2:04E 04 6:38E 05 1:33E 05 7:71E 05 0 0 0
101.33 1.49 2.14 79
101.33 1.38 1.90 79
159 2:72E 07 2:68E 07 ± 6:93E 05 2:30E 04 7:16E 05 1:42E 05 8:57E 05 6 12 11
101.33 1.71 0.92 80
101.33 1.52 2.32 55
134 1:52E 07 1:39E 07 1:15E 07 4:43E 05 1:19E 04 4:55E 05 1:46E 05 6:01E 05 10 29 22
512.50 1.37 2.41 79
101.33 1.59 3.16 61
133 1:70E 07 1:61E 07 7:80E 06 4:31E 05 1:38E 04 4:45E 05 1:45E 05 5:90E 05 9 30 23
184.50 1.65 0.99 72
184.30 1.46 2.11 59
140 1:32E 07 1:23E 07 1:24E 07 3:39E 05 1:04E 04 3:49E 05 1:50E 05 4:99E 05 13 45 35
101.33 1.48 2.14 81
512.00 1.27 2.48 57
136 1:79E 07 1:56E 07 1:19E 07 4:54E 05 1:34E 04 4:68E 05 1:69E 05 6:37E 05 27 27 17
101.33 1.72 0.93 79
101.33 1.75 3.05 116 181 1:61E 07 1:57E 07 ± 4:10E 05 1:34E 04 4:23E 05 1:35E 05 5:58E 05 1 34 28
101.33 1.27 0.67 65
101.33 1.29 1.87 137 196 1:15E 07 1:11E 07 8:36E 06 3:37E 05 9:50E 03 3:46E 05 1:44E 05 4:90E 05 8 46 36
454.50 1.17 0.76 59
376.50 1.26 1.59 128 193 1:17E 07 8:85E 06 9:11E 06 3:43E 05 7:57E 03 3:51E 05 1:45E 05 4:95E 05 8 45 36
101.33 1.25 0.67 65
Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2
D
If the top product of column 1 (AB) is taken and fed to column 2 as vapor then TAC ($/yr) is 7:51E 05.
Total actual plates Heating rate (kJ/h) Cooling rate (kJ/h) Main HX duty (kJ/h) Steam cost ($/yr) C.W cost ($/yr) Operat. cost ($/yr) Capital cost ($/yr) TAC ($/yr) Capital savings (%) Op. cost savings (%) TAC savings (%)
Pressure (kPa) Diameter (m) Re¯ux ratio Actual plates
Description
Table 9 Case 2 optimal schemes based on US utility prices
1312 M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
a
178 2:41E 07 2:36E 07 ± 1:64E 06 8:20E 04 1:72E 06 1:42E 05 1:86E 06 0 0 0
DQF
IQF
DQB
IQB
SP
SQF
SQB
101.33 1.52 2.32 55
146 1:51E 07 1:38E 07 1:15E 07 1:31E 06 4:80E 04 1:36E 06 1:50E 05 1:51E 06 6 21 19
512.50 1.36 2.39 91
101.33 1.39 2.13 60
148 1:69E 07 1:60E 07 1:15E 07 1:15E 06 5:56E 04 1:21E 06 1:63E 05 1:37E 06 15 30 26
202.50 1.61 0.99 88
184.30 1.46 2.11 59
149 1:32E 07 1:21E 07 1:24E 07 9:00E 05 4:19E 04 9:42E 05 1:54E 05 1:10E 06 9 45 41
101.33 1.48 2.12 90
512.00 1.26 2.47 60
152 1:78E 07 1:55E 07 1:19E 07 1:21E 06 5:38E 04 1:26E 06 1:76E 05 1:44E 06 24 27 23
101.33 1.71 0.91 92
101.33 1.73 2.99 132 207 1:59E 07 1:54E 07 ± 1:08E 06 5:36E 04 1:13E 06 1:45E 05 1:28E 06 2 34 31
101.33 1.26 0.65 75
101.33 1.29 1.84 151 212 1:14E 07 1:10E 07 8:27E 06 9:96E 05 3:82E 04 1:03E 06 1:49E 05 1:18E 06 5 40 36
455.00 1.16 0.76 61
377.00 1.25 1.57 137 202 1:17E 07 8:79E 06 9:05E 06 1:02E 06 3:05E 04 1:05E 06 1:47E 05 1:19E 06 3 39 36
101.33 1.25 0.65 65
Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 101.33 1.38 1.87 93
183 2:70E 07 2:66E 07 ± 1:84E 06 9:22E 04 1:93E 06 1:52E 05 2:08E 06 7 12 12
101.33 1.71 0.91 90
Col. 1
101.33 1.49 1.79 90
Col. 2
Col. 1
101.33 1.48 2.12 88
Ia
D
If the top product of column 1 (AB) is taken and fed to column 2 as vapor then TAC ($/yr) is 1:79E 06.
Total actual plates Heating rate (kJ/h) Cooling rate (kJ/h) Main HX duty (kJ/h) Steam cost ($/yr) C.W cost ($/yr) Operat. cost ($/yr) Capital cost ($/yr) TAC ($/yr) Capital savings (%) Op. cost savings (%) TAC savings (%)
Pressure (kPa) Diameter (m) Re¯ux ratio Actual plates
Description
Table 10 Case 2 optimal schemes based on European utility prices
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317 1313
1314
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
Table 11 Winning structures in Case 2 (equimolar feed) USA prices
EU prices
Structure
Savings (%)
Structure
Savings (%)
SQF SQB DQB IQF DQF
36 36 35 23 22
DQB SQF SQB SP IQF
41 36 36 31 26
savings) and the third ranked backward integrated sloppy system (SQB, 30% savings) involve unpreferable integration AB ! C or the even wider boiling point gap A ! C comparing to the B ! BC integration in the winner structure. Case 3: This is the case where the energy integrated sloppy structures, namely the ®rst ranked backward energy integrated sloppy system (SQB, 53% savings) and the second ranked forward integrated sloppy system (SQF, 51% savings) come to play since the heat duty is really high. The Petlyuk system (PS) with its 30% savings is just fourth behind the third ranked direct sequence with backward integration (DQB, 30% savings). The preference of the energy integration over the direct column coupling is supported by controllability studies. Degrees of freedom analysis and steady-state multivariable control structure synthesis tools show [4] that the investigated schemes can be controlled by conventional decentralised control structures, but in case of the energy integration the interaction among the control loops is less than in case of the Petlyuk system. Steady state and dynamic controllability studies [7] show that both the energy-integrated and the thermally coupled structures can be operated. However, the control of the Petlyuk system is signi®cantly poorer. Comparing the optimisation results of rigorous simulations to the corollaries of the short-cut analysis, the domain where the thermally coupled system wins is even much lower, and is constrained to a small area near the AC edge of the composition triangle at balanced relative volatility ratio. At balanced relative volatility ratio and equimolar feed composition, as well as near the node of pure B, the Petlyuk system takes only the fourth place behind the energy-integrated structures at European price structures. SP is not ranked in the ®rst ®ve places in the case of the American price structure, in these feed compositions. Yet, the Petlyuk system wins over the energy-integrated structures at 10% middle component in the feed. Its advantage, in TACs, over the second best (energy-integrated) structure is no more than 5±6%. Considering all the energy, cost, operability, and ¯exibility viewpoints, the advantageous application of the thermally coupled systems, if indeed exists, is constrained to a very small range of relative volatility ratio, feed composition, and price structure. This small range is situated somewhere around balanced ratio relative volatility ratio A/B to B/C, small amount of the middle component B, balanced presence of the two swing components A and C in the feed or a little bit shifted to the direction of C. On the other hand, the integrated sloppy structures win, in a great TAC percent, at high B ratio in the feed, while the conventional energy integrated structures, DQB in our particular case, is the best choice at equimolar feed.
Total actual plates Heating rate (kJ/h) Cooling rate (kJ/h) Main HX duty (kJ/h) Steam cost ($/yr) C.W cost ($/yr) Operat. cost ($/yr) Capital cost ($/yr) TAC ($/yr) Capital savings (%) Op. cost savings (%) TAC savings (%)
Pressure (kPa) Diameter (m) Re¯ux ratio Actual plates
Description
189 3.00E 07 2.99E 07 ± 2.04E 06 1.04E 05 2.15E 06 1.65E 05 2.31E 06 0 0 0
Col. 2 101.33 1.38 8.34 94
190 3.09E 07 3.08E 07 ± 2.10E 06 1.07E 05 2.21E 06 1.68E 05 2.37E 06 2 3 3
101.33 1.94 0.79 96
Col. 1
Col. 2 101.33 1.88 0.88 96
Col. 1 101.33 1.41 8.74 93
I
D
Table 12 Case 3 optimal schemes based on European utility prices Col. 2 145.50 1.80 0.92 94
145 2.61E 07 1.24E 07 1.25E 07 1.35E 06 4.30E 04 1.40E 06 1.76E 05 1.57E 06 7 35 32
101.33 1.50 10.06 51
Col. 1
DQB Col. 2 101.33 1.99 17.15 116
185 2.05E 07 2.03E 07 ± 1.39E 06 7.05E 04 1.46E 06 1.49E 05 1.61E 06 10 32 30
101.33 1.13 0.69 69
Col. 1
SP
Col. 2 101.33 1.50 11.11 127
168 1.37E 07 1.36E 07 1.14E 07 9.31E 05 4.80E 04 9.79E 05 1.58E 05 1.14E 06 4 54 51
338.50 1.33 1.60 41
Col. 1
SQF
Col. 2 295.50 1.37 9.53 122 178 1.30E 07 1.09E 07 1.09E 07 8.84E 05 3.79E 04 9.21E 05 1.69E 05 1.09E 06 2 57 53
101.33 1.37 1.09 56
Col. 1
SQB
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317 1315
1316
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
Table 13 Winning structures in Case 3 (80% middle component) EU prices Structure
Savings (%)
SQB SQF DQB SP
53 51 32 30
Acknowledgements We acknowledge the ®nancial support provided by Hungarian Scienti®c Research Fund (OTKA) grants T016851 and 021302. Appendix A A.1. Sizing and costing of distillation columns For a given number of theoretical trays, HYSYS simulator calculates column diameter after converging for selected valve tray distillation column with 50.8-mm weir height. Valve trays of Glitsch type are considered. In order to estimate the actual number of trays, overall column ef®ciency is calculated using simpli®ed equation [4]: log Eo 1:67
0:25 log lavg aavg 0:30 log Lm =Vm 0:30 hl
Nactual Ntheoretical =Eo
A:1 A:2
The height of the column for 0.6 m tray spacing and 6 m disengagement is given by: H Nactual
1 0:6 6:0
A:3
where Nactual 1 0:6 tray stack height (m). The costing of distillation columns (carbon steel construction) can be estimated by the following cost equations that are updated from mid-1968 to 1997 using the ratio of Marshall & Swift index (1056.8/280). Installed cost of the column shell;
$ M&S=280 937:61D1:066 H 0:802 3:18
If the design pressure (P) is more than 345 kPa, a correction factor 1 1:45 10 4 P applied. Installed cost of column;
$ M&S=280 136:14D1:55 h
Total column cost Installed cost of column shell Installed cost of column trays
A:4 345 is A:5 A:6
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
1317
A.2. Sizing and costing of heat exchangers The heat transfer area (A) of heat exchangers was calculated as A Q= U LMTD
A:7
by assuming the following overall heat transfer coecients (U): · Ur 3400 kJ=m2 h °C for boilers, · Uc 2800 kJ=m2 h °C for condensers, · Ue 2100 kJ=m2 h °C for liquid/liquid heat exchangers. The cost of heat exchangers can be correlated as a function of the surface area. The prices were updated from mid-1968 to 1997 by M&S index (1056.8/280), assuming shell and tube, ¯oating head, and carbon steel construction. Installed cost of heat exchanger;
$ M&S=280 474:67A0:65 3:29
A:8
where: A area m2 ; 18:6 < A < 464:5 and design pressure up to 1034.2 kPa. A.3. Annual capital cost The capital cost (purchase plus installation cost) is annualised over a period which is often referred to as plant life time Annual capital cost Capital cost=Plant life time TAC Annual operating cost Annual capital cost
A:9 A:10
Operating costs were assumed just utility cost (steam and cooling water). · Plant lifetime 10 years · Operating hours 8000 h/year. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
O. Annakou, P. Mizsey, Ind. Engng. Chem. 35 (1996) 1877. Z. Fonyo, J. Szabo, P. Foldes, Acta Chim. 82 (1974) 235. G. Kaibel, Chem. Engng. Technol. 10 (1987) 92. P. Mizsey, N. Hau, N. Benko, I. Kalmar, Z. Fonyo, Comp. Chem. Engng. 22 (1998) S427. M.S. Peter, K.D. Timmerhaus, Plant Design and Economic for Chemical Engineer, third ed., Mc Graw-Hill, New York, 1988. F.B. Petlyuk, V.M. Platonov, D.M. Slavinskii, Int. Chem. Engng. 5 (3) (1965) 561. E. Rev, M. Emtir, Z. Szitkai, P. Mizsey, Z. Fony o, Comp. Chem. Engng. 25 (2001) 119±140. J. Stichlmair, A. Stemmer, Chem. Engng. Technol. 12 (1989) 163. W.J. Stupin, Ph.D. Dissertation, University of Southern California, 1970. R.O. Wright, US patent 2471,134, 1945.
www.elsevier.com/locate/apthermeng
Rigorous simulation of energy integrated and thermally coupled distillation schemes for ternary mixture Mansour Emtir, Endre Rev *, Zsolt Fonyo Department of Chemical Engineering, Budapest University of Technology and Economics, H-1521 Budapest, Hungary Received 28 November 2000; accepted 22 January 2001
Abstract Sharp (99% purity) separation of a ternary mixture, characterised by near uniformly distributed volatility, by direct separation sequence without, with forward, and with backward energy integration, by indirect separation sequence without, with forward, and with backward energy integration, by sloppy separation sequence with forward and with backward energy integration, and by thermally coupled sloppy separation sequence (Petlyuk system) is rigorously modelled and optimised. Three feed compositions, namely (Case 1) equimolar A/C ratio with 10% B, (Case 2) equimolar feed, and (Case 3) equimolar A/C ratio with 80% B are compared. Comparison is based on total annual costs (TACs) using European and American price systems. The savings in TAC of Petlyuk are uniformly about 28±33% in all the three cases, while the savings of the energy integrated systems increases together with increasing ratio of B in the feed. In Case 1, Petlyuk system is the winner, with 33% savings, but a conventional energy-integrated system is handicapped by just a very few percent. In Cases 2 and 3, Petlyuk system is not amongst the best structures. In Case 2, either the conventional energy-integrated systems or the energy-integrated sloppy structures win with 35±41%; while Case 3, the energy-integrated sloppy structures are the best with about 51% savings. Ó 2001 Elsevier Science Ltd. All rights reserved.
1. Introduction Distillation is the most widely used separation technique in the petrochemical and chemical process industries for the separation of ¯uid mixtures despite its high energy requirement. Signi®cant energy savings can be reached by the use of complex distillation arrangements such as the side-stripper, the side-recti®er, the thermal (internal) column coupling (also known as Petlyuk
*
Corresponding author. Fax: +36-1-463-3197. E-mail address: [email protected] (E. Rev).
1359-4311/01/$ - see front matter Ó 2001 Elsevier Science Ltd. All rights reserved. PII: S 1 3 5 9 - 4 3 1 1 ( 0 1 ) 0 0 0 1 7 - 5
1300
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
Nomenclature A A B C D D DQF DQB Eo ESI hW hl H I IQF IQB Lm
heat transfer area light component middle component heavy component column diameter (m) conventional direct sequence direct sequence with forward energy integration direct sequence with backward energy integration overall plate eciency ease of separation index, ESI aAB =aBC weir height (m) liquid height above valve opening (m) column height (m) conventional indirect sequence indirect sequence with forward energy integration indirect sequence with backward energy integration molar liquid ¯ow rate (kmol/h)
M&S N P Q S SQD SQB SQF SP T TAC U Vm x
Marshall and Swift index number of plates pressure heat duty (kJ/h) sloppy sequence without integration or coupling (Pre¯ash system) sloppy sequence with double energy integration sloppy sequence with backward energy integration sloppy sequence with forward energy integration thermally coupled sloppy sequence (Petlyuk system) temperature total annual costs overall heat transfer coecient molar vapor ¯ow rate (kmol/h) mole fraction
Greeks a Relative volatility k Average latent heat Average viscosity of feed (cP) lavg
system), the (external) energy integration (also known as energy integrated distillation system) and the heat pumping techniques. Theoretical studies, [1,2,6,8,9] have shown that the column coupling con®gurations are capable of achieving typically 30% of energy savings compared to a conventional sequence. In addition, the coupling con®guration can also be achieved with the so called dividing wall column, by placing a vertical wall in the middle of the column separating the feed from the side draw, e.g. [3,10]. With this arrangement, reduction in capital cost can also be expected through the elimination of one of the column shells (but not the column internal). Despite the above advantages, industry has been reluctant to use the Petlyuk system and dividing wall columns and this is usually attributed to the lack of established design procedures and the fear of control problems. This paper compares the energy integrated and thermally coupled con®gurations based on rigorous simulation and optimisation for total annual costs (TACs). In this article the following abbreviations are used for distillation structures, including not only those rigorously optimised but also those studied, by short-cut methodology, in our earlier publication [7]:
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
D DQF DQB I IQF IQB S SQF SQB SQD SP
1301
Direct separation sequence without energy integration (Fig. 1). Direct separation sequence with forward energy integration (Fig. 2). Direct separation sequence with backward energy integration (Fig. 3). Indirect separation sequence without energy integration (Fig. 4). Indirect separation sequence with forward energy integration (Fig. 5). Indirect separation sequence with backward energy integration (Fig. 6). Sloppy separation sequence without any energy integration or thermal coupling (Fig. 7). Sloppy separation sequence with forward energy integration (Fig. 8). Sloppy separation sequence with backward energy integration (Fig. 9). Sloppy separation sequence with double (both forward and backward) energy integration (Fig. 10). Thermally coupled sloppy separation sequence (Petlyuk system) (Fig. 11).
The sloppy separation with or without a single energy integration (S, SQF, and SQB) can be realised by three columns or with a system of a pre¯ash column and a side-stream column obtained by lumping the second and third columns. The double integrated sloppy structure SQD is
Fig. 1. Direct separation sequence without energy integration (D).
Fig. 2. Direct separation sequence with forward energy integration (DQF).
1302
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
Fig. 3. Direct separation sequence with backward energy integration (DQB).
Fig. 4. Indirect separation sequence without energy integration (I).
Fig. 5. Indirect separation sequence with forward energy integration (IQF).
realized by a three-column system. Lumping the two end columns is a precondition for realizing the Petlyuk structure SP.
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
1303
Fig. 6. Indirect separation sequence with backward energy integration (IQB).
Fig. 7. Sloppy separation sequence without any energy integration or thermal coupling (S).
According to our previously performed short-cut analysis [7], in the case of sharp separation the sloppy separation path with forward, backward, or double energy integration (SQF, SQB, or SQD) is almost always capable of achieving as much energy savings as that of with the thermal coupling (Petlyuk system or dividing wall column, SP). Moreover, the conventional direct sequence and indirect sequence structures with energy integration (DQF, DQB, IQF, IQB) are, in some cases, also capable of achieving larger energy savings than that of the corresponding Petlyuk or dividing wall column con®gurations (SP).
1304
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
Fig. 8. Sloppy separation sequence with forward energy integration (SQF).
Fig. 9. Sloppy separation sequence with backward energy integration (SQB).
The main corollaries of the short-cut analysis are the following: From energy point of view the integrated and coupled structures are always better than the non-integrated ones. From energy point of view the integrated or thermally coupled structures almost always win. There is just a very small domain for DQ to win at very high ratio of A in the feed. All the integrated and coupled sloppy structures are equivalent. Considering exergy loss, the integrated and coupled structures share the Gibbs composition triangle. There is a signi®cant area where SP wins near the AC edge. According to the short-cut analysis, the forward or backward energy-integrated, as well as the double energy-integrated sloppy structures (SQF, SQB, and SQD) are equivalent, in energy term,
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
1305
Fig. 10. Sloppy separation sequence with double (both forward and backward) energy integration (SQD).
Fig. 11. Thermally coupled sloppy separation sequence (Petlyuk system) (SP).
to the thermally coupled sloppy structures (PS, i.e. Petlyuk or dividing wall column con®guration); and the energy-integrated structures win, in energy term, almost everywhere of the studied conditions. These include a wide range of relative volatility ratios. The Petlyuk structure has the greatest chance to win over the energy-integrated schemes at balanced relative volatility ratio. Even in that case, the integrated structures win over the Petlyuk system in the greatest part of the feed composition triangle; but the Petlyuk system yet proves to being a winner in a signi®cant area of feed compositions.
1306
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
2. Rigorous case studies Continuous recti®cation of a ternary mixture of ethanol, n-propanol, and n-butanol at 99% product purity speci®cations and three dierent feed compositions are simulated and optimized in this study. The feed compositions, candidates for the SP con®guration to win, are selected according to the results of the short-cut analysis. The main data of the mixtures to be separated are listed in Tables 1±3.
Table 1 Feed and product speci®cations for Case 1 Streams
Feed
Ethanol
n-Propanol
n-Butanol
kmol/h
Fraction
kmol/h
Fraction
kmol/h
Fraction
kmol/h
Fraction
Components Ethanol (A) n-Propanol (B) n-Butanol (C)
135 30 135
0.45 0.1 0.45
134.87 1.36 0.00
0.99 0.01 0.00
0.14 27.27 0.14
0.005 0.990 0.005
0.00 1.36 134.87
0.00 0.01 0.99
Total
300
1.000
136.23
1.00
27.55
1.000
136.23
1.00
Table 2 Feed and product speci®cations for Case 2 Streams
Feed
Ethanol
kmol/h
Fraction
Components Ethanol (A) n-Propanol (B) n-Butanol (C)
100 100 100
0.333 0.333 0.333
Total
300
1.000
kmol/h
n-Propanol
n-Butanol
Fraction
kmol/h
Fraction
kmol/h
Fraction
99.50 1.01 0.00
0.99 0.01 0.00
0.49 98.00 0.49
0.005 0.990 0.005
0.00 1.01 99.50
0.00 0.01 0.99
100.51
1.00
98.98
1.000
100.51
1.00
Table 3 Feed and product speci®cations for Case 3 Streams
Feed
Ethanol
n-Propanol
n-Butanol
kmol/h
Fraction
kmol/h
Fraction
kmol/h
Fraction
kmol/h
Fraction
Components Ethanol (A) n-Propanol (B) n-Butanol (C)
30 240 30
0.10 0.80 0.10
28.79 0.29 0.00
0.99 0.01 0.00
1.21 239.42 1.21
0.005 0.990 0.005
0.00 1.01 99.50
0.00 0.01 0.99
Total
300
1.000
29.08
1.00
241.84
1.000
29.08
1.00
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
1307
Table 4 Relative volatilities and ease of separation indices aAB aAC aBC ESI
Case 1
Case 2
Case 3
2.02 4.67 2.31 0.87
2.07 4.72 2.29 0.90
2.15 4.79 2.23 0.96
Fig. 12. Composition triangle with the location of the feed compositions.
The relative volatilities are approximately balanced in this material system. The mean relative volatilities and estimated ease of separation are shown in Table 4. The feed points are located near the centre of the AC edge in Case 1, in the centre of the Gibbs composition triangle Case 2, and near the node of pure B in Case 3 (see Fig. 12). In either case, 99% product purity and equal distribution of the impurities is speci®ed in all the products. The D, DQF, DQB, I, IQF, IQB, S, SQF, SQB, and SP (Petlyuk) structures are simulated, optimized, and then compared. The neither integrated nor thermally coupled sloppy system (the so-called pre¯ash system) with three columns (S) and the double integrated sloppy structure (SQD) are not studied rigorously; the ®rst one because of its evident inferiority, the second one because of its expected excess complexity in realization. The costs of the dividing wall column (a variant of SP) is, on the other hand, not determined; the Petlyuk system is studied, instead. The two conventional systems D and I are studied just for obtaining a base case to determine the savings. HYSYS simulation package is applied for rigorous modelling in all studied systems with NRTL thermodynamic property model using the data sets built in the simulator, atmospheric feed and products at saturated liquid state. Two radically dierent cost structures are taken into account. One of them is a high utility price structure corresponding to prices in Europe; the other one is a low utility price structure that corresponds to prices in the USA. Utility cost data are collected in Table 5. The eect of the high and low utility costs is demonstrated in Cases 1 and 2; therefore, Case 3 is studied with the European cost structure only. Valve trays (of Glitsch type) are considered as column internal. Installation cost formulas [5] and data are collected in Appendix A.
1308
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
Table 5 Utilities cost data Utility LP-steam MP-steam Cooling water Electricity a b
High utility pricesa
Low utility pricesb
Temperature level (°C)
Price
Temperature level (°C)
Price
160 184 30±45 ±
17.7 $/ton 21.8 $/ton 0.027 $/ton 0.1 $/kW h
160 184 30±45 ±
6.62 $/ton 7.31 $/ton 0.0067 $/ton 0.06 $/kW h
Based on European prices. Based on USA prices.
Results of Case 1 are collected in Tables 6 and 7. Here the feed is situated near the AC edge. The presence of component B in the feed is just 10%; the presence of components A and B are equimolar to each other. The two tables correspond to the two subcases of American and European utility prices. The winners and their rate of winning are collected in Table 8. SP wins, in TAC savings compared to the base case D, with 4% and 3% above DQB, and with 7% and 6% above IQF. The Petlyuk system is better than the others in both of its lower operating costs and capital costs are. According to the short-cut analysis, at this feed composition the Petlyuk structure (SP) ought to win over one of the integrated structures with approximately 40%. According to Tables 6 and 7, however, in case of the lower utility costs (American price structure) the handicap of DQB behind SP is just 5.7% in TAC, 8% in capital costs, and 4.8% in energy costs. In case of the higher utility costs (European price structure), the handicap of DQB behind SP is again no more than 4.85% in TAC, 7.8% in capital costs, and 5.55% in energy costs. Results of Case 2 are collected in Tables 9 and 10. Here the feed is equimolar in the components. The winners and their rate of winning are collected in Table 11. According to the short-cut analysis, at this feed composition the Petlyuk structure (SP) ought to win over one of the integrated structures with approximately 20%. On the contrary, SP does not win in any of the two rigorous subcases. We also investigated these cases with sieve trays, and got the same qualitative results (not listed here). Instead of SP, the integrated sloppy structures SQF and SQB win in the case of American utility prices. However, they are just 1% better above DQB, in TAC savings compared to the base case D. According to Tables 9 and 10, in case of the lower utility costs (American price structure) the handicap of DQB behind SQF and SQB is just 1.8% and 0.8% in TAC, 4.1% and 3.5% in capital costs, and 0.87% and 0.57% in energy costs. (In the latest ®gure, DQB is better in energy costs over SQB.) In case of the higher utility costs (European price structure) the structures change place. Here DQB wins over the sloppy structures. The handicap of SQF and SQB behind DQB is 7.3% and 8.2% in TAC, and 9.3% and 11.5% in energy costs, while they have better capital cost ®gures with 3.2% and 4.5%. The Petlyuk system (SP) is not in the ®rst ®ve places in the case of the American price structure; and it is just at the 4th place, with great handicap, in the case of the higher (European) utility prices.
a
155 1:98E 07 1:93E 07 ± 5:04E 05 1:65E 04 5:21E 05 1:19E 05 6:40E 05 0 0 0
DQF
IQF
DQB
IQB
SP
SQF
SQB
101.33 1.35 7.73 47
130 1:53E 07 1:36E 07 8:04E 06 4:47E 05 1:17E 04 4:58E 05 1:38E 05 5:97E 05 16 12 7
512.50 1.39 1.50 83
101.33 1.36 1.30 53
124 1:37E 07 1:21E 07 1:14E 07 3:49E 05 1:03E 04 3:59E 05 1:45E 05 5:04E 05 22 31 21
249.00 1.42 0.92 71
230.50 1.20 6.36 50
128 1:33E 07 1:19E 07 8:01E 06 3:39E 05 1:01E 04 3:49E 05 1:35E 05 4:84E 05 13 33 24
101.33 1.49 1.26 78
512.00 1.23 1.62 52
123 1:38E 07 1:19E 07 1:22E 07 3:51E 05 1:02E 04 3:61E 05 1:50E 05 5:11E 05 26 31 20
101.33 1.51 0.85 71
101.33 1.54 1.32 126
187 1:27E 07 1:22E 07 ± 3:23E 05 1:04E 04 3:33E 05 1:25E 05 4:58E 05 5 36 28
101.33 1.31 0.64 61
101.33 1.24 0.94 114 166 1:22E 07 1:17E 07 7:15E 06 3:57E 05 9:98E 03 3:67E 05 1:37E 05 5:04E 05 15 30 21
495.00 1.22 0.74 52
458.00 1.28 1.06 105 158 1:27E 07 9:21E 06 9:66E 06 3:70E 05 7:88E 03 3:78E 05 1:43E 05 5:21E 05 20 27 19
101.33 1.31 0.63 53
Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 101.33 1.34 1.10 85
156 2:34E 07 2:29E 07 ± 5:95E 05 1:96E 04 6:15E 05 1:28E 05 7:43E 05 8 18 16
101.33 1.51 0.85 71
Col. 1
101.33 1.21 5.46 75
Col. 2
Col. 1
101.33 1.49 1.26 80
Ia
D
If the top product of column 1 (AB) is taken and fed to column 2 as vapor then TAC ($/yr) is 6.04E 05.
Total actual plates Heating rate (kJ/h) Cooling rate (kJ/h) Main HX duty (kJ/h) Steam cost ($/yr) C.W cost ($/yr) Operat. cost ($/yr) Capital cost ($/yr) TAC ($/yr) Capital savings (%) Op. cost savings (%) TAC savings (%)
Pressure (kPa) Diameter (m) Re¯ux ratio Actual plates
Description
Table 6 Case 1 optimal schemes based on US utility prices
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317 1309
a
IQF
DQB
IQB
SP
SQF
SQB
176 2:32E 07 2:27E 07 ± 1:58E 06 7:87E 04 1:66E 06 1:37E 05 1:79E 06 12 18 17
101.33 1.34 1.09 93
101.33 1.35 7.73 47
143 1:52E 07 1:36E 07 8:04E 06 1:32E 06 4:70E 04 1:37E 06 1:43E 05 1:51E 06 16 3 1
512.50 1.37 1.49 96
101.33 1.36 1.30 53
137 1:36E 07 1:21E 07 1:14E 07 9:24E 05 4:19E 04 9:66E 05 1:50E 05 1:12E 06 22 31 27
249.50 1.41 0.90 84
230.50 1.20 6.36 50
136 1:32E 07 1:18E 07 8:01E 06 9:01E 05 4:09E 04 9:42E 05 1:38E 05 1:08E 06 12 33 30
101.33 1.48 1.24 86
512.00 1.23 1.62 52
135 1:37E 07 1:17E 07 1:22E 07 9:30E 05 4:07E 04 9:70E 05 1:55E 05 1:13E 06 27 31 27
101.33 1.51 0.82 83
101.33 1.53 1.31 128
195 1:26E 07 1:21E 07 ± 8:59E 05 4:20E 04 9:01E 05 1:28E 05 1:03E 06 4 36 33
101.33 1.30 0.63 67
101.33 1.24 0.93 116 172 1:22E 07 1:16E 07 7:09E 06 1:06E 06 4:04E 04 1:10E 06 1:39E 05 1:24E 06 13 22 19
495.00 1.21 0.73 56
457.00 1.28 1.06 105 160 1:27E 07 9:20E 06 9:64E 06 1:10E 06 3:19E 04 1:13E 06 1:43E 05 1:28E 06 17 19 17
101.33 1.31 0.63 55
Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2
DQF
If the top product of column 1 (AB) is taken and fed to column 2 as vapor then TAC ($/yr) is 1:42E 06.
166 1:97E 07 1:92E 07 ± 1:34E 06 6:67E 04 1:41E 06 1:22E 05 1:53E 06 0 0 0
101.33 1.51 0.82 83
Col. 1
101.33 1.20 5.39 86
Col. 2
Col. 1
101.33 1.49 1.26 80
Ia
D
Total actual plates Heating rate (kJ/h) Cooling rate (kJ/h) Main HX duty (kJ/h) Steam cost ($/yr) C.W cost ($/yr) Operat. cost ($/yr) Capital cost ($/yr) TAC ($/yr) Capital savings (%) Op. cost savings (%) TAC savings (%)
Pressure (kPa) Diameter (m) Re¯ux ratio Actual plates
Description
Table 7 Case 1 optimal schemes based on European utility prices
1310 M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
1311
Table 8 Winning structures in Case 1 (10% middle component) USA prices
EU prices
Structure
Savings (%)
Structure
Savings (%)
SP DQB IQF SQF IQB
28 24 21 21 20
SP DQB IQF IQB SQF
33 30 27 27 19
Results of Case 3 are collected in Table 12. Here the feed is situated near the B node. The presence of component B in the feed is 80%; the presence of components A and B are equimolar to each other. Just the subcase of the European utility prices is shown. As the ratio of component B increases in the feed, the energy demand of the separation increases, and the energy costs become dominating. The structures DQF, IQF, and IQB are proved inferior to the others, as the energy costs are dominant; therefore, they are omitted. The conventional indirect structure I is, however included for comparison. The winners and their rate of winning are collected in Table 13. In accordance with the results of the short-cut analysis, this is the place where the integrated sloppy structures (SQB and SQF) win. The Petlyuk system (SP) is again at the fourth place behind the third rank DQB. The handicap of the DQB system behind SQB and SQF is 44% and 38% in TAC, 4.1% and 5.7% in capital costs, and 52% and 43% in energy costs.
3. Conclusions A general rule governing the ranks of the studied structures according to TAC is the increasing heat duty requirement with increasing concentration of component B in the feed. The structures with energy integration are sensitive for the heat duty while the Petlyuk system without energy integration (but thermal coupling) is not. The savings in TAC of Petlyuk are uniformly about 30± 33% in all the three cases. Case 1: Here the Petlyuk system (PS) is the winner with 28% or 33% savings. There is no qualitative ranking dierence considering the two price structures. Second ranked is the direct sequence with backward energy integration (DQB) with 24% or 30% savings that cannot be more because of low heat duty. It is back just with 3%. The forward energy integrated sloppy system (SQF, 19% savings) is just the third or fourth ranked, probably because there is an AB ! C energy integration that involves a pressure shift and use of high pressure steam. Case 2, American prices: Here the Petlyuk system is not amongst the best structures. Direct sequence with backward energy integration (DQB) with 35% savings is at the third place backed just with 1% by both forward energy integrated sloppy system (SQF) and backward energy integrated sloppy system (SQB) by their 36% savings. Case 2, European prices: Here the direct sequence with backward integration (DQB) is the winner with 41% savings. Both the second ranked forward integrated sloppy system (SQF, 36%
a
Ia DQF
IQF
DQB
IQB
SP
SQF
SQB
101.33 1.50 1.82 77
156 2:43E 07 2:39E 07 ± 6:17E 05 2:04E 04 6:38E 05 1:33E 05 7:71E 05 0 0 0
101.33 1.49 2.14 79
101.33 1.38 1.90 79
159 2:72E 07 2:68E 07 ± 6:93E 05 2:30E 04 7:16E 05 1:42E 05 8:57E 05 6 12 11
101.33 1.71 0.92 80
101.33 1.52 2.32 55
134 1:52E 07 1:39E 07 1:15E 07 4:43E 05 1:19E 04 4:55E 05 1:46E 05 6:01E 05 10 29 22
512.50 1.37 2.41 79
101.33 1.59 3.16 61
133 1:70E 07 1:61E 07 7:80E 06 4:31E 05 1:38E 04 4:45E 05 1:45E 05 5:90E 05 9 30 23
184.50 1.65 0.99 72
184.30 1.46 2.11 59
140 1:32E 07 1:23E 07 1:24E 07 3:39E 05 1:04E 04 3:49E 05 1:50E 05 4:99E 05 13 45 35
101.33 1.48 2.14 81
512.00 1.27 2.48 57
136 1:79E 07 1:56E 07 1:19E 07 4:54E 05 1:34E 04 4:68E 05 1:69E 05 6:37E 05 27 27 17
101.33 1.72 0.93 79
101.33 1.75 3.05 116 181 1:61E 07 1:57E 07 ± 4:10E 05 1:34E 04 4:23E 05 1:35E 05 5:58E 05 1 34 28
101.33 1.27 0.67 65
101.33 1.29 1.87 137 196 1:15E 07 1:11E 07 8:36E 06 3:37E 05 9:50E 03 3:46E 05 1:44E 05 4:90E 05 8 46 36
454.50 1.17 0.76 59
376.50 1.26 1.59 128 193 1:17E 07 8:85E 06 9:11E 06 3:43E 05 7:57E 03 3:51E 05 1:45E 05 4:95E 05 8 45 36
101.33 1.25 0.67 65
Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2
D
If the top product of column 1 (AB) is taken and fed to column 2 as vapor then TAC ($/yr) is 7:51E 05.
Total actual plates Heating rate (kJ/h) Cooling rate (kJ/h) Main HX duty (kJ/h) Steam cost ($/yr) C.W cost ($/yr) Operat. cost ($/yr) Capital cost ($/yr) TAC ($/yr) Capital savings (%) Op. cost savings (%) TAC savings (%)
Pressure (kPa) Diameter (m) Re¯ux ratio Actual plates
Description
Table 9 Case 2 optimal schemes based on US utility prices
1312 M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
a
178 2:41E 07 2:36E 07 ± 1:64E 06 8:20E 04 1:72E 06 1:42E 05 1:86E 06 0 0 0
DQF
IQF
DQB
IQB
SP
SQF
SQB
101.33 1.52 2.32 55
146 1:51E 07 1:38E 07 1:15E 07 1:31E 06 4:80E 04 1:36E 06 1:50E 05 1:51E 06 6 21 19
512.50 1.36 2.39 91
101.33 1.39 2.13 60
148 1:69E 07 1:60E 07 1:15E 07 1:15E 06 5:56E 04 1:21E 06 1:63E 05 1:37E 06 15 30 26
202.50 1.61 0.99 88
184.30 1.46 2.11 59
149 1:32E 07 1:21E 07 1:24E 07 9:00E 05 4:19E 04 9:42E 05 1:54E 05 1:10E 06 9 45 41
101.33 1.48 2.12 90
512.00 1.26 2.47 60
152 1:78E 07 1:55E 07 1:19E 07 1:21E 06 5:38E 04 1:26E 06 1:76E 05 1:44E 06 24 27 23
101.33 1.71 0.91 92
101.33 1.73 2.99 132 207 1:59E 07 1:54E 07 ± 1:08E 06 5:36E 04 1:13E 06 1:45E 05 1:28E 06 2 34 31
101.33 1.26 0.65 75
101.33 1.29 1.84 151 212 1:14E 07 1:10E 07 8:27E 06 9:96E 05 3:82E 04 1:03E 06 1:49E 05 1:18E 06 5 40 36
455.00 1.16 0.76 61
377.00 1.25 1.57 137 202 1:17E 07 8:79E 06 9:05E 06 1:02E 06 3:05E 04 1:05E 06 1:47E 05 1:19E 06 3 39 36
101.33 1.25 0.65 65
Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 Col. 1 Col. 2 101.33 1.38 1.87 93
183 2:70E 07 2:66E 07 ± 1:84E 06 9:22E 04 1:93E 06 1:52E 05 2:08E 06 7 12 12
101.33 1.71 0.91 90
Col. 1
101.33 1.49 1.79 90
Col. 2
Col. 1
101.33 1.48 2.12 88
Ia
D
If the top product of column 1 (AB) is taken and fed to column 2 as vapor then TAC ($/yr) is 1:79E 06.
Total actual plates Heating rate (kJ/h) Cooling rate (kJ/h) Main HX duty (kJ/h) Steam cost ($/yr) C.W cost ($/yr) Operat. cost ($/yr) Capital cost ($/yr) TAC ($/yr) Capital savings (%) Op. cost savings (%) TAC savings (%)
Pressure (kPa) Diameter (m) Re¯ux ratio Actual plates
Description
Table 10 Case 2 optimal schemes based on European utility prices
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317 1313
1314
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
Table 11 Winning structures in Case 2 (equimolar feed) USA prices
EU prices
Structure
Savings (%)
Structure
Savings (%)
SQF SQB DQB IQF DQF
36 36 35 23 22
DQB SQF SQB SP IQF
41 36 36 31 26
savings) and the third ranked backward integrated sloppy system (SQB, 30% savings) involve unpreferable integration AB ! C or the even wider boiling point gap A ! C comparing to the B ! BC integration in the winner structure. Case 3: This is the case where the energy integrated sloppy structures, namely the ®rst ranked backward energy integrated sloppy system (SQB, 53% savings) and the second ranked forward integrated sloppy system (SQF, 51% savings) come to play since the heat duty is really high. The Petlyuk system (PS) with its 30% savings is just fourth behind the third ranked direct sequence with backward integration (DQB, 30% savings). The preference of the energy integration over the direct column coupling is supported by controllability studies. Degrees of freedom analysis and steady-state multivariable control structure synthesis tools show [4] that the investigated schemes can be controlled by conventional decentralised control structures, but in case of the energy integration the interaction among the control loops is less than in case of the Petlyuk system. Steady state and dynamic controllability studies [7] show that both the energy-integrated and the thermally coupled structures can be operated. However, the control of the Petlyuk system is signi®cantly poorer. Comparing the optimisation results of rigorous simulations to the corollaries of the short-cut analysis, the domain where the thermally coupled system wins is even much lower, and is constrained to a small area near the AC edge of the composition triangle at balanced relative volatility ratio. At balanced relative volatility ratio and equimolar feed composition, as well as near the node of pure B, the Petlyuk system takes only the fourth place behind the energy-integrated structures at European price structures. SP is not ranked in the ®rst ®ve places in the case of the American price structure, in these feed compositions. Yet, the Petlyuk system wins over the energy-integrated structures at 10% middle component in the feed. Its advantage, in TACs, over the second best (energy-integrated) structure is no more than 5±6%. Considering all the energy, cost, operability, and ¯exibility viewpoints, the advantageous application of the thermally coupled systems, if indeed exists, is constrained to a very small range of relative volatility ratio, feed composition, and price structure. This small range is situated somewhere around balanced ratio relative volatility ratio A/B to B/C, small amount of the middle component B, balanced presence of the two swing components A and C in the feed or a little bit shifted to the direction of C. On the other hand, the integrated sloppy structures win, in a great TAC percent, at high B ratio in the feed, while the conventional energy integrated structures, DQB in our particular case, is the best choice at equimolar feed.
Total actual plates Heating rate (kJ/h) Cooling rate (kJ/h) Main HX duty (kJ/h) Steam cost ($/yr) C.W cost ($/yr) Operat. cost ($/yr) Capital cost ($/yr) TAC ($/yr) Capital savings (%) Op. cost savings (%) TAC savings (%)
Pressure (kPa) Diameter (m) Re¯ux ratio Actual plates
Description
189 3.00E 07 2.99E 07 ± 2.04E 06 1.04E 05 2.15E 06 1.65E 05 2.31E 06 0 0 0
Col. 2 101.33 1.38 8.34 94
190 3.09E 07 3.08E 07 ± 2.10E 06 1.07E 05 2.21E 06 1.68E 05 2.37E 06 2 3 3
101.33 1.94 0.79 96
Col. 1
Col. 2 101.33 1.88 0.88 96
Col. 1 101.33 1.41 8.74 93
I
D
Table 12 Case 3 optimal schemes based on European utility prices Col. 2 145.50 1.80 0.92 94
145 2.61E 07 1.24E 07 1.25E 07 1.35E 06 4.30E 04 1.40E 06 1.76E 05 1.57E 06 7 35 32
101.33 1.50 10.06 51
Col. 1
DQB Col. 2 101.33 1.99 17.15 116
185 2.05E 07 2.03E 07 ± 1.39E 06 7.05E 04 1.46E 06 1.49E 05 1.61E 06 10 32 30
101.33 1.13 0.69 69
Col. 1
SP
Col. 2 101.33 1.50 11.11 127
168 1.37E 07 1.36E 07 1.14E 07 9.31E 05 4.80E 04 9.79E 05 1.58E 05 1.14E 06 4 54 51
338.50 1.33 1.60 41
Col. 1
SQF
Col. 2 295.50 1.37 9.53 122 178 1.30E 07 1.09E 07 1.09E 07 8.84E 05 3.79E 04 9.21E 05 1.69E 05 1.09E 06 2 57 53
101.33 1.37 1.09 56
Col. 1
SQB
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317 1315
1316
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
Table 13 Winning structures in Case 3 (80% middle component) EU prices Structure
Savings (%)
SQB SQF DQB SP
53 51 32 30
Acknowledgements We acknowledge the ®nancial support provided by Hungarian Scienti®c Research Fund (OTKA) grants T016851 and 021302. Appendix A A.1. Sizing and costing of distillation columns For a given number of theoretical trays, HYSYS simulator calculates column diameter after converging for selected valve tray distillation column with 50.8-mm weir height. Valve trays of Glitsch type are considered. In order to estimate the actual number of trays, overall column ef®ciency is calculated using simpli®ed equation [4]: log Eo 1:67
0:25 log lavg aavg 0:30 log Lm =Vm 0:30 hl
Nactual Ntheoretical =Eo
A:1 A:2
The height of the column for 0.6 m tray spacing and 6 m disengagement is given by: H Nactual
1 0:6 6:0
A:3
where Nactual 1 0:6 tray stack height (m). The costing of distillation columns (carbon steel construction) can be estimated by the following cost equations that are updated from mid-1968 to 1997 using the ratio of Marshall & Swift index (1056.8/280). Installed cost of the column shell;
$ M&S=280 937:61D1:066 H 0:802 3:18
If the design pressure (P) is more than 345 kPa, a correction factor 1 1:45 10 4 P applied. Installed cost of column;
$ M&S=280 136:14D1:55 h
Total column cost Installed cost of column shell Installed cost of column trays
A:4 345 is A:5 A:6
M. Emtir et al. / Applied Thermal Engineering 21 (2001) 1299±1317
1317
A.2. Sizing and costing of heat exchangers The heat transfer area (A) of heat exchangers was calculated as A Q= U LMTD
A:7
by assuming the following overall heat transfer coecients (U): · Ur 3400 kJ=m2 h °C for boilers, · Uc 2800 kJ=m2 h °C for condensers, · Ue 2100 kJ=m2 h °C for liquid/liquid heat exchangers. The cost of heat exchangers can be correlated as a function of the surface area. The prices were updated from mid-1968 to 1997 by M&S index (1056.8/280), assuming shell and tube, ¯oating head, and carbon steel construction. Installed cost of heat exchanger;
$ M&S=280 474:67A0:65 3:29
A:8
where: A area m2 ; 18:6 < A < 464:5 and design pressure up to 1034.2 kPa. A.3. Annual capital cost The capital cost (purchase plus installation cost) is annualised over a period which is often referred to as plant life time Annual capital cost Capital cost=Plant life time TAC Annual operating cost Annual capital cost
A:9 A:10
Operating costs were assumed just utility cost (steam and cooling water). · Plant lifetime 10 years · Operating hours 8000 h/year. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
O. Annakou, P. Mizsey, Ind. Engng. Chem. 35 (1996) 1877. Z. Fonyo, J. Szabo, P. Foldes, Acta Chim. 82 (1974) 235. G. Kaibel, Chem. Engng. Technol. 10 (1987) 92. P. Mizsey, N. Hau, N. Benko, I. Kalmar, Z. Fonyo, Comp. Chem. Engng. 22 (1998) S427. M.S. Peter, K.D. Timmerhaus, Plant Design and Economic for Chemical Engineer, third ed., Mc Graw-Hill, New York, 1988. F.B. Petlyuk, V.M. Platonov, D.M. Slavinskii, Int. Chem. Engng. 5 (3) (1965) 561. E. Rev, M. Emtir, Z. Szitkai, P. Mizsey, Z. Fony o, Comp. Chem. Engng. 25 (2001) 119±140. J. Stichlmair, A. Stemmer, Chem. Engng. Technol. 12 (1989) 163. W.J. Stupin, Ph.D. Dissertation, University of Southern California, 1970. R.O. Wright, US patent 2471,134, 1945.