352
Ind. Eng. Chem. Fundam. 1985, 2 4 , 352-359
Incentives for Dual Composition Control in Single and Heat- Integrated Binary Distillation Columns Teh-ping Chlang and Wllllam L. Luyben' Department of Chemical Engineering, Lehlgh Universiw, Bethlehem, Pennsylvania 180 75
Minimum energy consumption is achieved in a distillation cdumn if all product compositions are controlled. However, instrumentation complexity is increased and the potential for control loop interaction must be considered. Steady-state calculations were employed to evaluate the incentives for controlling all product compositions in conventional single columns and several heat-Integrated designs. Energy savings when controtling all compositions were compared with simple one-end control. The results of this study indicate that in many binary distlktbn systems there is very l i incentive for controlling the compositions of all product streams, even in heat-integrated systems. This study was limited to binary systems, both ideal and real, over a range of relative volatilities, symmetrical product purities, and feed compositions. These conclusions are in no way intended to imply that analysis of most distillation columns is unnecessary. Virtually every column should be given an initial screening to determine which control configuration works best and to see whether more detailed analysis is justified.
Introduction Due to increasing fuel costs, control of distillation columns has been intensively studied. Luyben (1975) has shown that minimum energy consumption is achieved when both product compositions are controlled. Such a scheme has been called dual composition control or twopoint control; see Figure 1. Although dual composition control results in minimum energy usage, interaction between the two composition loops may generate closed-loop stability difficulties (Rijnsdorp, 1965). Also, instrumentation complexity and engineering costa are significantly increased if dynamic simulation studies are necessary for control system design. Many academic studies of dual composition control have appeared in the literature (McAvoy and Weischedel, 1980, 1981; Gagnepain and Seborg, 1982; Waller, 1982). The principal objective has been to minimize the interaction between the composition loops. Little attention has been paid to the topic of finding the economic incentives of dual composition control (Luyben, 1975). Simple, stable, oneend control schemes include fixed reflux-to-feed, steamto-feed, and reflux-to-distillate flow ratios. Overfractionation is used deliberately, and energy consumption is increased as a consequence. This study extends the work of Luyben (1975) to examine the steady-state energy conservation aspects of a number of columns including several heat-integrated systems. Both ideal and real binary systems were studied, covering a range of relative volatilities, product purities, and feed compositions. Three specific chemical systems were studied: methanol-water, benzene-toluene, and isobutane-n-butane. Single Columns A. Systems Studied. 1. Ideal Systems. Columns with relative volatility (a)from 1.2 to 7.0 were studied on the basis of the following assumptions: (a) binary system, (b) equimolal overflow, (c) theoretical trays, (d) total condenser, (e) partial reboiler, (f)actual reflux ratio is 1.2 times the minimum, and (g) saturated liquid feed. Product purities ranged from 95% to 99.9%. Only symmetrical separations were considered; i.e., impurity levels in both products were equal (X,= 1 - XD). This assumption was made for several reasons. First, it greatly reduces the number of cases that would have to be studied. Second, it is fairly close to the typical situation in many 0196-43131a5/ 1024-03528015010
industrial columns. Third, unsymmetrical product compositions have been explored by Luyben (1975). Columns were explored for a range of feed compositions from 30% and 80%. 2. Real Systems. Three important industrial distillation systems that are typical of many commercial separations were studied: methanol-water, benzene-toluene, and isobutane-n-butane. The assumptions made were negligible pressure drop through the column and tray efficiency of 75%. Columns were designed as follows. The minimum reflux ratio was found by extending the vapol-liquid equilibrium (VLE) line through the feed point on the enthalpy-composition diagram to the intersection with a vertical line through the distillate composition. The actual reflux ratio was set a t 1.2 times the minimum. The number of trays was determined by plate-to-plate calculations from the specified bottoms composition to the specified distillate purity. Specifications of the columns studied are listed in Table I. B. Single-Column Control System Design. Steady-state calculations were employed to evaluate the incentives for several single binary distillation columns. The energy consumptions of four alternative control schemes are compared: constant product composition, constant reflux flow rate, constant heat input, and constant reflux ratio. Feed composition changes of f10% (on an absolute basis) were considered. Comparions were made for different values of relative volatility a,product purities XD and Xg,and feed composition XF. C. Control Systems. 1. Constant Product Composition Case (See Figure 1). The vapor boilup V and reflux R required to achieve the specified product purities XD and XB were calculated as feed compositions varied from the design case. These calculations were performed for a fixed column, i.e. one with a fixed number of total trays NT and a constant feed tray location NF. Three curves were generated: reflux-to-feed ratio R / F , vapor boilup-to-feed ratio V / F ,and reflux-to-distillate ratio RR. Figure 2 gives an example of these curves for a system with XD= 0.999, XB = 0.001,XF = 0.3 at design, NF = 23, NT = 41, and a = 2.0. 2. Constant Reflux Case (SeeFigure 3). The reflux flow rate is fixed at the maximum value obtained above, and either product composition is held constant by manipulating vapor boilup. Since excess reflux is used, the 0 1985 American Chemical Society
Ind. Eng. Chem. Fundam., Vol. 24, No. 3, 1985
.I.,&
353
&.----
1
2 .
Figure 3. Constant reflux operation.
Figure 1. Dual composition control.
V/ F
.
R /F . ..
Figure 4. Constant vapor boilup operation. .h5
.3
.35
.h
*F
Figure 2. Variations of system variables during perfect control of product compositions.
uncontrolled product purity will always exceed specification. For the example given above, a t the design point XF = 0.3, the V / F ratio is 0.025 higher than the minimum dual composition control where XD is held constant. This represents an average increase in energy consumption of only 1.7%. 3. Constant Vapor Boilup Case (See Figure 4). Vapor boilup is fixed a t the maximum value obtained in section 1 above, and either product composition is held constant by manipulating reflux. Since excess vapor boilup is being used, the composition of the uncontrolled product will always exceed specification. At XF= 0.3, by using the
same example as above, the V/F ratio is 0.117 higher than dual composition control, which represents a 8% increase in energy consumption. 4. Constant Reflux Ratio (See Figure 5). The reflux ratio is fixed a t the maximum value obtained in section 1above, and either product composition is held constant by manipulating vapor boilup. Since both excess vapor boilup and reflux are being used, the composition of the uncontrolled product will always be better than specification For the example, the reflux ratio is 2.07 higher than for dual composition control at XF = 0.3, which represents a 45% increase in energy consumption. In order to quantify the energy wasted by non-dualcomposition control, a term called the “additional energy percentage” (AEP) is defined. It is the energy difference between the non-dual-composition and the dual compo-
354
Ind. Eng. Chem. Fundam., Vol. 24, No. 3, 1985
Table I. Specifications for Single Columns Studied (F= 2300 g-mol/h) system XB/XD P, mmHg TF, "C methanol-water
0.001/0.999
760
57.0
benzene-toluene
0.001 /0.999
300
57.0
isobutane-n- butane
0.02/0.98
5060
30.0
XF 0.3 0.5 0.8 0.3 0.5 0.8 0.3 0.5 0.8
NF 10 9 11 24 21 19 42 43 43
NT
38 42 42 42 40 42 80 80 73
104QB, cal/h 14.1 20.8 30.4 21.2 24.2 27.6 43.6 45.6 44.6
Table 11. Summary of AEP for the Ideal Systems Studied a
XB/XD
7.0
0.001/0.999
5.0
0.001/0.999 0.02/0.98
2.0
0.001/0.999 0.02/0.98
1.2
0.02/0.98 0.05/0.95
AEP, % const V 22.5 15.4 8.8
const RR 11.0 5.0 2.5
XF 0.3 0.5 0.8
NT 15 15 16
( V/F)design
0.493 0.7 1.0
const R 2.3 1.1 1.0
0.3 0.5 0.8 0.3 0.5 0.8
18 18 20 11 11 9
0.593 0.79 1.09 0.554 0.75 0.99
2.9 1.6 0.1 0.1 1.5 7.6
19.7 14.0 8.0 18.2 12.0 3.0
24.5 9.4 3.5 30.0 12.0 11.7
0.3 0.5 0.8 0.3 0.5 0.8
41 40 42 23 24 21
1.463 1.675 1.98 1.41 1.6 1.81
1.7 1.5 0.1 1.6 0.7 7.5
45.0 20.0 12.0 44.0 21.0 17.0
0.3 0.5 0.8 0.3 0.5 0.8
85 87 80 64 67 59
5.91 6.19 6.17 5.41 5.82 5.23
8.0 7.5 4.0 9.2 5.0 1.7 4.1 0.8 5.5 8.0 0.34 16.0
2.4 0.7 7.3 6.2 1.4 18.0
51.0 26.0 21.0 60.0 30.0 25.0
Table 111. AEP of Constant Reflux Ratio Cases ( X , = 0.8) a XdX, N T AEP, % 1.2 2.0 2.0 5.0 5.0 7.0
0.02/0.98 0.02/0.98 0.001/0.999 0.02/0.98 0.001/0.999 0.001/0.999
80 21 42 9 20 16
21.0 17.0 12.0 11.7 3.5 2.3
sition control systems as a percentage of the design heat input. D. Case Study Results. 1. Ideal Cases. The "AEP's" of the distillation systems studied are given in Table 11. XD is held constant for all these systems. The control strategy, from among the three alternatives to dual composition control (constant R, constant V , or constant RR), that gives the lowest AEP value is the best alternative. An examination of Table I1 shows some interesting facts. First, most of these cases show little incentive for dual composition control since the lowest AEP is small (less than 5%). Second, low relative volatility systems with high feed compositions are found to have incentives for dual composition control. Third, the constant reflux ratio scheme shows high AEP in most of these cases. Table I11 gives the general trend of this scheme for different CY'Sand purities. It appears that the reflux ratio scheme becomes beneficial in systems with very high CY and high purities. This result is seen in the methanol-water system to be discussed later. Fourth, none of these one-end control schemes is the best in all cases. The constant reflux scheme uses less energy than the other schemes in most cases, but not in all. No general conclusions could be drawn as to what is the best control system from an energy conservation point of
r-@
Aia
Figure 5. Constant reflux ratio operation.
view for all the distillation columns. Steady-state calculations considering all four alternative control systems should be made for each particular design problem. All one needs is a steady-state column program. The VIF, RIF, and RR curves are generated as feed composition varies over a typical operating range in the plant. The curve that shows the least change should be evaluated as
Ind. Eng. Chem. Fundam., Vol. 24, No. 3, 1985 355 Table IV. Summary of
AEP for the Real Systems Studied XBJXD O.ool/0.999
system methanol-water
XF
NT
0.3
38 42 42 42 40 42 80 80 73
AEP, % const QB 26.4 16.4 7.3 6.6 6.6 3.2 1.35 0.1 6.6
const R 13.7 7.7 1.4 2.4 0.7 0.5 0.92 1.9 8.8
const RR 5.1 1.7 2.76 31.0 13.0 7.0
1
25 3
>
c
a
0
2w =. .
=>
a y
m 020-
uo w ’ c
a
17c
Figure 7. Feed-split configuration.
n
E6 A H
Y
O .4
XF
Figure 6. Variations of system variables during perfect control of product compositions (methanol-water separation).
an alternative to dual composition control. 2. Real Cases. Table IV gives the AEP’s of the three real systems studied. Since benzene-toluene and isobutane-n-butane are fairly ideal systems, the results shown are similar to the ideal cases studied. Methanol-water is a nonideal, high relative volatility system. Figure 6 shows results for the example where XF = 0.5 at design. The very flat RR curve gives the constant reflux ratio operation an AEP of 1.7%. The constant reflux ratio scheme is the best a t XF = 0.3 and XF = 0.5, and it is comparable to the constant reflux scheme a t XF = 0.8. This confirmed the conclusion drawn from the ideal case study that the constant reflux ratio scheme is an alternative to dual composition control for high-cu systems. Heat-Integrated Columns Heat-integrated distillation columns are an important component for energy efficient process design. A heatintegrated distillation system is composed of two or more thermally linked columns. One column operates a t a high enough pressure so that the overhead vapor can be used as a heating medium for another column. Chiang and Luyben (1983) studied five different heat-integrated configurations for the methanol-water separation. They concluded that three configurations had fairly similar advantages over a conventional single column. That work
1 Y I
I
t t l Figure 8. Light split/reverse configuration.
I
has been extended to study several other chemical systems including both ideal and real cases. The feed-split (FS) configuration (see Figure 7) consumed the least amount of energy for the isobutane-n-butane separation. The light split/reverse (LS/R)configuration (see Figure 8) was the best scheme for the methanol-water separation. In this study, these two configurations were used to explore the incentives for controlling all product compositions in heat-integrated distillation columns. “All composition control” of heat-integrated distillation columns was studied by extending the method used for the single-column case. Simple control schemes, such as controlling one end in the first column and dual composition control on the other (1-2 scheme) or controlling one end in each column (1-1 scheme), are compared with con-
356
Ind. Eng. Chem. Fundam., Vol. 24, No. 3, 1985
Table V. Specifications of the Isobutane-n -Butane Separation (FS Configuration) XF, mole fraction 0.3 col 1 1117 0.98 0.02 10850 107 57 26.2
feed rate, g-mol/h distillate comp (mole fraction isobutane) bottoms comp (mole fraction isobutane) operating pressure, mmHg no. of trays feed tray location reboiler heat duty, IOfi cal/h
0.5 col 2 1183 0.98 0.02 5060
col 1 1119 0.98 0.02 10850 107 58 27.1
ao
42 22.4
0.8
col 2 1181 0.98 0.02 5060
col 1 1118 0.98 0.02 10850 96 58 26.4
ao
43 23.4
col 2 1182 0.98 0.02 5060 73 43 22.9
Table VI. FS 2-2 Scheme (All Composition Control) for Isobutane-n -Butane Separation x p
0.2 0.25 03" 0.35 0.4 (i.4
0.45 0.50 9.55 o.6 0.7 0.75 0.8" 0.85 0.b a
RRH
F H/ F L
4.765 4.903 4.984 5.025 5.037
25.4 20.5 i7.1 14.6 12.7
0.946 0.945 0.9448 0.9452 0.9458
26.9 27.1 27.12 27.14 27.1
5.043 5.031 5.0 4.949 4.88
12.7 11.2 10.0 8.9 8.1
0.946 0.947 0.948 0.949 0.95
28.2 27.5 26.4 2.5.0 22.7
5.1 4.872 4.576 4.177 3.58
7.2 6.4 5.6 4.8 3.9
0.935 0.94 0.946 0.954 0.967
104QBH, cal/h 24.8 25.6 26.2 26.6 26.9
RHIFH
RRL
RLIFL 3.592 3.695 3.755 3.785 3.793 3.799 3.788 3.762 3.722 3.667 3.764 3.597 3.381 3.087 2.648
19.2 15.4 12.9 11.0 9.6 9.6 8.5 7.5 6.7 6.1 5.3 4.7 4.2 3.6 2.9
Design point.
Figure 9. 2-2 scheme (FS)
trolling both ends in each column (2-2 scheme) from the standpoint of steady-state energy consumption. A. Control System Design of FS Configuration. The specific system studied was the isobutane-n-butane separation at three design feed compositions: 30%, 50%, and 80%. Specifications of these columns are listed in Table V. 1. 2-2 Scheme (See Figure 9). Four compositions (composition of bottoms and distillate products in each column) were controlled. As feed composition varied from the design case, the heat input to the high-pressure column (QBH), reflux to the high-pressure column (RH), reflux to the low-pressure column (RL),the feed-split ratio ( F H I F L ) , and the reflux ratio of each column (RRH, RRL) were calculated such that the specified product purities were achieved; see Figure 10. Those calculations were performed for a fixed system, i.e. one with fixed NTH, N F H , NTL,and NFL. These rating calculations follow the following sequence: (1) Guess initial feed flow rate to the high-pressure column F H . (2) Calculate QBH,QDH, RH, and RRH required to produce specification products in the high-pressure column. (3) Set heat input QBL to the lowpressure column equal to 95% of the condenser heat duty of the high-pressure column, i.e. 95% of QDH. (5% heat
0 4
a
0
P
5 10
4
a
U.
%
15-
-I Y
9 5.5
4.5 12-1
3.5
Figure 10. FS variations of system variables for 2-2 scheme (isobutane-n-butane separation).
lost was assumed.) (4)Set feed flow rate to the lowpressure column F L = F - F H . (5) Calculate QBL, RL, and RRL required to produce specification products in the low-pressure column. (6) If QBL is equal to the value set by step 3, the correct value of FH has been determined.
Ind. Eng. Chem. Fundam., Vol. 24, No. 3, 1985 357
Table VII. Specifications of the Methanol-Water Separation (LS/R Configuration) Xr, mole fraction 0.3 0.5 col 1 col 2 col 1 col 2 2300 1948 2300 1718 feed rate, g-mol/h 0.999 0.999 0.999 0.999 distillate comp (mole fraction methanol) 0.174 0.331 0.001 0.001 bottoms comp (mole fraction methanol) 3500 760 760 2800 operating pressure, mmHg 53 42 49 no. of trays 38 4 13 18 feed tray location 6 7.4 10.5 12.7 10.0 reboiler heat duty, lo6 cal/h
0.8 col 1 2300 0.999 0.662 760 42 4 15.4
col 2 1358 0.999 0.001 2280 51 11 17.3
L_ - _ _ _ -J _ _ - _ - _ _ _ .
Figure 11. 2-1 scheme (FS).
Otherwise reguess F H and go to step 2. Table VI shows the QBH, R H / F H , RRH,FH/FL, RLIFL, and RRL values as feed composition was varied *lo%. As shown in the single-column studies, the variables with the least amount of change indicate what variables should be held constant for more simple control strategies. Figure 10 plots an example where the design feed composition is 30%. The FH/FL,RH/FH,and RL/FLcurves are flat. These flat curves indicate that there may be little incentive for the complex “all composition” (2-2) control system. A more simple 2-1 scheme or 1-1 scheme may waste very little energy. The following discussions are based on the 30% feed composition example. 2. 2-1 Scheme (SeeFigure 11). Three compositions (XDH, X B H in the high-pressure column, and X D L in the low-pressure column) are controlled. The FH/FLratio is fixed a t the maximum value obtained in the 2-2 calculations. Since excess heat is used in the low-pressure column, the uncontrolled product purity will always exceed specification. The rating procedures used above can be easily extended to this case. The feed rate loop iteration is not necessary, and in step 5 Q B L and X D L have been specified so that R L and the uncontrolled X B L can be calculated. By use of the same example as above, F H / F L is set a t 0.946, which is 0.001 higher than the minimum dual composition (2-2) control. The AEP is only 0.04%. 3. 1-1 Scheme (SeeFigure 12). The composition of one product in each column is controlled. Excess reflux is used in both columns so that the uncontrolled product compositions will always exceed specifications. R H is fixed a t the maximum value obtained in the 2-2 calculations, and the corresponding FH/FL ratio is used as the feed-split ratio. If one extends the method used in the 2-2 scheme calculations, the design procedure is as follows: (1) R H , FH/FL, and XDHare fixed. QBH and XBHcan be calculated in the high-pressure column. (2) Set QBL = 0 . 9 5 Q ~(5% ~ heat lost was assumed.) (3) Q B L and X D L are fixed. R L and X B L can then be calculated in the low-pressure column. For the example given above, R H is fixed a t the maximum which occurs a t XF = 0.4. The corresponding feedsplit ratio is 0 . 9 4 5 8 . Distillate compositions of each column
Figure 12. 1-1 scheme (FS).
Figure 13. 1-2 scheme (LS/R).
are held constant. The AEP is only 1.0%. Table IX gives results for all the FS cases for the isobutane-n-butane separation. The only case which shows any significant incentive for any control system more complex than the simple 1-1 system is the high feed composition case (6.8% AEP) where a 2-1 system appears to be justified. These results indicate that most FS heat-integrated columns do not require dual composition control on both columns. B. Control System Design of LS/R Configuration. The specific system studied was the methanol-water separation with design feed compositions of 30%, 50%,and 80%. Specifications of these columns are listed in Table VII. 1. 1-2 Scheme (SeeFigure 13). Three compositions were controlled for this configuration: X D L in the lowpressure column and X D H and X B H in the high-pressure column. The QBH, R H , RRH, R L , and RRL were calculated as feed composition was varied around the design case. Those calculations were performed for fixed columns. The calculation procedure was as follows: (1)Guess the bottoms composition in the low-pressure column ( X B L ) . (2) Calculate QBL, R L , and RRL required to produce the specified separation in the low-pressure column. (3) Set FH= BL and T m = TBp (4) Calculate QBH, QDH, RH, and
358
Ind. Eng. Chem. Fundam., Vol. 24, No. 3, 1985
Table VIII. LS/R 1-2 Scheme (All Composition Control) for Methanol-Water Separation XF
XBL
0.2 0.25 0.3" 0.35 0.4
0.11 0.14 0.17 0.20 0.24
RLIFL 0.126 0.15 0.177 0.204 0.23
RRL 1.25 1.18 1.14 1.12 1.09
10*QBH, cal/h 8.1 9.0 10.0 11.0 12.0
0.4 0.45 0 3 0.55 0.6
0.25 0.29 0.33 0.38 0.42
0.204 0.227 0.249 0.271 0.292
1.01 1.o 0.98 0.97 0.95
0.7 0.75 0.8" 0.85 0.9
0.53 0.59 0.66 0.73 0.82
0.328 0.344 0.357 0.367 0.369
0.92 0.9 0.87 0.84 0.8
RH/FH
RRH
0.253 0.291 0.338 0.391 0.448
2.32 2.08 1.97 1.91 1.87
11.0 11.9 12.7 13.6 14.5
0.376 0.424 0.476 0.531 0.589
1.51 1.47 1.44 1.41 1.39
15.8 16.6 17.3 17.9 18.3
0.688 0.751 0.814 0.872 0.916
1.29 1.26 1.23 1.18 1.12
Design point.
Table IX. Summary of AEP for the Heat-Integrated Systems Studied XB/XD XF FS isobutane-n-butane 0.02/0.98 0.3 0.3 0.5 0.5 0.8 0.8 LS/R
'
methanol-water
0.3 0.5 0.5 0.8 0.8
0.001/0.999
AEP, %
control strategy 2-1 1-1 2-1 1-1 2-1 1-1
fixed variables FH/Fr
F H / F L I QBH
0.04 1.0 0.08 1.0 1.4 6.8
1-1 1-1
RRH RRH RRH, RRL RH R H , RL
8.1 1.3 1.7 0.9 1.4
0-1 1-1 0- 1
FHIFL, R H FHIFL F H / F L p QBH FH/FL
5j Figure 15. 1-1 scheme (LS/R).
0
0
.7
B XF
.9
.7
.a
.9
XF
Figure 14. LS/R variations of system variables for 1-2 scheme (methanol-water separation).
RRH required to produce the specified separation in the high-pressure column. (5) If 95% of QDH was equal to QBL (5% heat lost was assumed), the correct value of XBLhad been determined. Otherwise XBLwas reguessed and we went back to step 2. Table VlII shows the QBH, R H / F H , RRH, RLIFL, and RRL values as feed composition varied from design. The variables which were fixed for more simple control strategies are given in Table IX for the cases studied. Figure 14 shows an example where XF = 0.8 a t design. The flat curves exhibited in RH/FH,RL/FL,RRH, and RRLindicate that a simple 1-1 scheme or a 0-1 scheme may use very
Figure 16. 0-1 scheme (LS/R).
little additional energy compared to the more complex 1-2 control scheme. The following discussions are based on the 80% feed composition example. 2. 1-1 Scheme (See Figure 15). R H was fixed a t the maximum value obtained in the 1-2 scheme calculations.
Ind. Eng. Chem. Fundam., Vol. 24, No. 3, 1985 359
Distillate composition in the high-pressure column was held constant by manipulating QBH. Since excess reflux was used, the uncontrolled product purity will always exceed specification. The AEP was only 0.9%. 3. 0-1 Scheme (See Figure 16). Only one composition was controlled. X D H was controlled by manipulating QBH. Excess reflux was used in both columns. Since R L and RH showed less variation with feed composition change in the 1-2 case, RL and R H were fixed a t the maximum values obtained in 1-2 scheme calculations, which occurred at X, = 0.9 in both columns. Only X D H was held constant. The AEP was only 1.4%. Table IX summarizes the AEP's of all the LS/R heatintegrated systems studied. For the LS/R configuration with methanol-water there appears to be little incentive for the complex 2-1 control system except a t low feed compositions. In fact in most cases, a very simple 0-1 scheme wastes very little energy. If we compare the results for the heat-integrated columns with the single-column cases presented in the first part of this paper, we find the same trends. Columns that showed little incentive for dual composition control in single-column designs also showed little incentive for "all composition control" in the heat-integrated design. It should be noted that this study has been limited to binary systems. Some preliminary work on multicomponent systems has indicated that they may show much higher incentives for dual composition control. A future paper will deal with this subject. McAvoy, in a recent paper (Stanley and McAvoy, 1984), has shown that the incentives for dual composition control increase if the dynamic aspects are considered.
alternative control scheme can also be determined. Systems with low relative volatility and high feed composition appear to be the only situation where the energy consumption of a simple control strategy is significantly higher (10%)than the dual composition control strategy for binary systems. The constant reflux scheme appears to be the best for most systems, except for the very high relative volatilities where the constant reflux ratio scheme is the best. The results for single-column systems can be extended to heat-integrated systems. Nomenclature AEP = additional energy percentage, % F = total feed flow rate, g-mol/h FS = feed-split configuration LS/R = light split/reverse configuration NF = feed tray location NT = total number of trays Q B = reboiler heat duty, lo6 cal/h QD = condenser heat duty, lo6 cal/h R = reflux flow rate, g-mol/h RR = reflux ratio TF = feed temperature, "C TB = bottoms temperature, "C V = vapor boilup rate, g-mol/h XB = bottoms composition, mole fraction XD = distillate composition, mole fraction XF = feed composition, mole fraction a = relative volatility
Conclusions Controlling the compositions of all products leaving a distillation column gives the minimum rate of energy consumption. However, control loop interaction and instrumentation complexity are increased. The more simple and stable control strategies, such as constant reflux, constant vapor boilup, or constant reflux ratio, often consume very little additional energy. These less complex schemes should be explored for any specific design problem. By computing the steady-state changes of all the system variables as feed composition varies over typical operating ranges in the plant, one can decide whether there are incentives for controlling all compositions. The best
Chiang, T. P.; Luyben, W. L. I n d . Eng. Chem. Process D e s . D e v . 1983, 22, 175. Gagnepaln, J. P.; Seborg, D. E. Ind. Eng. Chem. Process Des. Dev. 1982, 21, 5. Luyben, W. L. Ind. Eng. Chem. Fundam. 1975, 14, 321. McAvoy, T. J. AIChE J . 1981, 2 7 , 613. McAvoy, T. J.; Weischedel. K. Ind. Eng. Chem. Fundam. 1980, 19, 379. Rijnsdorp, J. E. Automatlka 1965, 1 , 15. Stanley, G. T.; McAvoy, T. J., paper presented at the course on Distillation Control, Bethlehem, PA, May 1984. Waller, K. V. In "Chemical Process Control"; Seborg. D. E., Edgar, R. P., Eds.; Engineerlng Foundatlon/American Institute of Chemical Engineers: New York, 1982; Vol. 2.
Subscripts H = high-pressure column L = low-pressure column Literature Cited
Received for review January 24, 1984 Revised manuscript received October 24, 1984 Accepted December 17, 1984
