Alternative distillation configurations for separating ternary mixtures

Alternative distillation configurations for separating ternary mixtures with small concentrations of intermediate in the feed. Imad M. Alatiqi, and Wi...
1 downloads 0 Views 799KB Size
500

Ind. Eng. Chem. Process Des. Dev. 1985,2 4 , 500-506

carbon mixtures to two independent properties. In particular, molecular weight is related to boiling point and specific gravity by the two equations cited below: (a) the equation proposed by Riazi and Daubert (1980)

M = a(Tb,R)b(s)C

M = (Tb/A)B (9) where M = molecular weight of the component boiling at temperature Tb (K) of the distillation curve 2.04 (TM)' 1 A = 22.31 +

(7)

S

with u = 4.5673 X b = 2.1962, and c = -1.0164, and (b) the equation recommended by the American Petroleum Institute (1983) M = 2.0438 X lo2 eXp(0.00218Tb,R) exp(-3.07S)(Tb,R)0'118(S)1'88... (8)

B = 1.27 +

0.086(TM)1'3 S

TM= molal average boiling point of the mixture, K, and s = specific gravity of the mixture. TM,the molal average boding point (K), can be estimated from Tv, the volumetric average boiling point (K), from the following relation TM = Tv - 5.85(S)'.33 (12)

Figure 2a compares the variation of molecular weight with boiling point for a value of specific gravity of 0.8 estimated according to eq 3, 7, and 8 with that obtained from the Winn nomograph. Figure 2b is a similar plot for a value of specific gravity of 0.9. It is seen that in both cases the results obtained with eq 3 conform most closely to the Winn nomograph. Finally, Table I1 presents a few examples comparing the molecular weights of some pure hydrocarbons with boiling points ranging from about 100 to 550 O C computed according to eq 3,7, and 8 with the formula weights of the compounds. The boiling point and the specific gravity of each of these compounds were taken from the work of Ferris (1955). It may be pointed out that both eq 7 and 8 use boiling temperature and specific gravity as the independent parameters to which molecular weight is related. Equation 3, on the other hand, uses boiling temperature and the Watson K factor as the independent parameters. Equation 3 is therefore more reliable for determining the variation of molecular weight with the extent of evaporation of an undefined mixture of hydrocarbons with the attendant unknown variation in specific gravity.

where S = average slope of the distillation curve, 10% to 90%; Co/liq. vol. %. Equation 12 is derived from a graph provided by Watson and Nelson (1933). 2. Equation 3 presented here appears to predict the molecular weight by hydrocarbons more accurately than eq 7 presented by Riazi and Daubert. It is likely that a correlation similar to eq 3 may serve as a better predictor for other properties too (with M representing the property under consideration) than the general equation proposed by those authors in the form of eq 7. Literature Cited American Petroleum Institute: Technical Data Book-Petroleum Refining, 4th Edition. 1983; pp 2-13. Brldgeman, 0. C. Research Paper RP694, National Bureau of Standards, US. Department of Commerce, 1934. Ferris, S. W. "Handbook of Hydrocarbons"; Academic Press Inc.: New Ywk, 1955. Hougen, 0. A,; Watson, K. M. "Industrial Chemical Calculatlons"; Wlley: New York, 1947. Lucy, F. A. Ind. Eng. Chem. 1938, 3 0 , 959. Malr, 6. J.; Willingham, C. 6. I d . Eng. Chem. 1936, 28, 1452. Rlazl, M. R.; Daubert, T. E. Hydrocarbon Process. 1980, 59, 115. Watson, K. M.; Nelson, E. F. Ind. Eng. Chem. 1933, 25, 880. Watson, K. M.; Nelson, E. F.; Murphy, G. B. Ind. Eng. Chem. 1935, 2 7 , 1460.

Conclusions

1. The method presented here provides a simple and quick procedure for estimating the molecular weight of any fraction of a multicomponent hydrocarbon blend, given the distillation curve and specific gravity of the original mixture. The relevant equations have been recast in the following form to facilitate their use.

Department of Mechanical Engineering V. Kuppu Rao Royal Military College of Canada Michael F. Bardon* Kingston, Ontario K7L 2W3 Received for review December 27, 1983 Accepted June 15, 1984

Alternative Distillation Conflguratlons for Separating Ternary Mixtures with Small Concentratlons of Intermediate in the Feed Single sidestream columns are shown to be inefficient for removing small amounts of the intermediate-boiling component In a ternary mixture. However, a sidestream column coupled with a sidestream stripper is shown to consume up to 30% less energy than the conventional two-column system.

sidestream column configuration. Tedder studied a range of feed compositions, using higher product purities (9&99%), and several configurations, including a sidestream column/stripper configuration. Both of these workers reported significantly lower energy consumption using nonconventional, complex configurations for some ranges of feed compositions. Elaahi and Luyben (1983) have explored four-component systems. A complex configuration, using two prefractionators and a two-sidestream third column, was shown to consume 25-35% less energy than the conven-

Introduction

Reduction of energy consumption in distillation has been actively studied in recent years as energy prices have increased sharply. Three-component systems have been explored by Doukas and Luyben (1978) and Tedder and Rudd (1978). Doukas studied systems where the feed had a small amount of either the lightest or the heaviest component. His study assumed low product purities (90-95%). He explored conventional two-column configurations as well as a simple sidestream column and a prefractionator/ 0196-4305/05/1124-0500$01.50/0

0

1985 American Chemical Society

Ind. [email protected]. Process Des. Dev., Vol. 24, No. 2, 1985 501

.. 14-

I--& XD(1) * .95 XD(2) = .05 XD(3) = .O

XD2(1) = 05 XDZl31 = 05

XD1121 = 05

I

XS(1) = .05 XS2) = .w XS(3) = .05

I

1 XB211) = 0

XB(1) = .o XW2) = .05 XB(3) = 95

XBZ(2) = 05 XBZ(3) = 95

Figure 1. LOF confiiation.

Figure 3. SS configuration.

.........

I

I

XD(2) * .05 XD(3) = .W

I I XS2 (1) = .05 x s 2 (2) = .eo XS2 (3) = .05 F

u XB1 (1) = ,005

1

Figure 2.

XB2 (1) =

.o

X82 (2) = .05 XB2 13) = 95

PF configuration.

0

XW1) XW2)

XW3)

tional three-column sequence. The purpose of this study was to extend Doukas’ work to the case where the concentration of the intermediateboiling component in the feed is low. The initial thought was that a single sidestream column might be the most economical configuration in this situation. The intermediate could be removed as a sidedraw purge from the column. Systems Studied. A ternary mixture of benzene, toluene, and 0-xylene was used in all cases. Column pressures were aesbned to be atmospheric. Four configurations of column sequences were explored. (1)“Light Out First” (LOF) configuration is the conventional two-column scheme where the lightest component is taken out of the top of the first column. The intermediate and heaviest components are separated in the second column. See Figure 1. The number of trays in each column was determined from the traditional Fenske-Underwood short-cut method using 1.1 times the minimum reflux ratio. Rigorous steady-state simulation was then used to get precise values for energy consumption. (2) “Prefractionator” (PF)confiiation makes a rough separation in the first column followed by the fiial separation into three products streams in a twefeed sidestream column. See Figure 2. Trial and error procedures were used to obtain the optimum design of this complex system. Reflux ratio, number of trays, and feed and sidestream tray locations were all varied until the minimum energy combination was achieved.

I

.oo

* .05 = .05

Figure 4. SSS configuration.

(3) “Single Sidestream” (SS) column consisted of only one column with a sidestream product. See Figure 3. Doukas showed that this configuration is the most economical when the feed contains less than 10% of either the lightest or the heaviest component. The sidedraw is withdrawn as a liquid from a tray above the feed tray if small amounts of the lightest component are in the feed. If small amounts of the heaviest component are in the feed, the sidedraw is withdrawn asea vapor from a tray below the feed tray. (4) ”Sidestream Stripper” (SSS)configuration uses a sidestream column with the liquid sidedraw fed into a small stripping column. The stripper has a small reboiler which removes some of the lightest component from the sidestream product. Vapor from the stripper is fed back into the main column. See Figure 4. Both the SS and SSS confiiations were optimized by trial and error procedures. The parameters included the liquid flow rate to the sidestream stripper (IS) sidestream , tray location (NS), and the number of trays in the stripper (NTS). The criterion of minimum energy requirement was taken as the primary basis for comparison between the different configurations. Two reasons were behind this choice: (i) Petlyuk et al. (1965) suggested that this criterion (EV/F) is adequate if the temperature levels of reboiler steam and cooling water are similar in all separation schemes. (ii) D o h and Luyben (1978) showed that even for complex configurations, utility costa constitute about 80% of the

502

Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 2, 1985

total cost of all systems studied. Equipment cost tends to scale almost directly with energy cost, e.g., column diameter, reboiler area, etc. Some brief consideration of capital cost will be discussed later. Numerical Techniques. The tridiagonal, bubble-point method of Wang and Henke (1967) was used to solve the algebraic equations describing the steady-state distillation columns. Convergence became difficult in some cases and the method had to be modified. These problems occurred when the concentration of the intermediate in the feed was very low. Such feed concentrations are sometimes referred to as “dumb-bell”feeds and have been reported by Seader (1981) to cause convergence difficulties when bubble-point methods are used. Convergence was improved by using the weighting factor proposed by Burningham and Otto (1967). The temperature on each tray for the k + 1iteration (Tk+J is changed by some fraction (WJ of the difference between the current bubble-point temperature (TBp) and the temperature from the previous iteration ( T k ) . Tk+i = Tk Wt(TBP - Tk) Values of Wt between 0.2 and 0.4 were used. The solutions for the columns in all configurations were sequential, except for the SSS configuration. The presence of flows in both directions between the main column and the stripper required an iterative procedure. Holland (1981) describes a “Capital 8 method” for convergence of such systems. For this work we used the original 8 method for each column and direct substitution for converging the whole system. The vapor rate and composition from the stripper were guessed and the main column was converged. Now the composition of the liquid stream to the stripper was known, so the stripper could be converged. The ratio of stripper vapor rate to the sidedraw rate was varied until the desired sidestream product composition was obtained. Then the new values for stripper vapor flow rate and composition were fed back into the main column, and it was converged again. The method converged very effectively in 3 to 5 cycles. Some interesting effects were observed when converging the stripper. The purity of the bottom product from the stripper is a nonmonotonic function of the vapor rate in the stripper. For low vapor rates the concentration of the lightest component is high so purity is low. As vapor rate is increased, the lightest component is stripped out of the product so purity increases. However, at still higher vapor rates, intermediate component begins to be removed in the vapor. Since the heaviest component is not stripped, its concentration begins to build up. Therefore, at high vapor rates, the purity of the stripper product (concentration of intermediate component) decreases as vapor rate is increased. Economic Basis. In this study the cost for each scheme is assumed to be the sum of utility cost and equipment annual cost (depreciation and maintenance) and is assumed to be 20% of total capital cost of distillation columns and heat exchangers. The basic column and heat exchanger costs were obtained from the graphs of Dryden and Furlow (1966) scaled by a factor of 3.28. Other costs such as piping, insulation, and instrumentation, were assumed to add 60% to the basic capital cost. Operating costs were assumed to be only utility costs (steam and cooling water). Cooling water cost was assumed to be 12~:/10000 lb. Steam at 300 psia was assumed to be $5.0/1000 lb and a t 200 psia, $4.0/1000 lb. The highpressure steam was used for process temperatures (TB) above 345 OF. The low-pressure steam was used for TB less than 345 O F .

Table I. SS Configuration sidestream energy consumption, purity, XS(2), mol 9L lo6 Btu/h

sidestream product flow rate LS, mol/h

40.0

6.74 7.60 8.70 10.17 12.29

21.8 15.7 12.6 10.7 19

18

l i

-

16

5 3

15

m

-

0

14

m

0

13

12

,

11

70

8

05

015

10

0 20

ZF(2i

Figure 5. Effect of feed composition on reboiler duty (low-purity products). (Purity of other product streams = 95%).

Since cost numbers are variable, detailed process design information is given in the tables to permit reevaluation of the various schemes using different cost numbers. Column diameters were determined by assuming a maximum superficial velocity of 2.5 ft/s. Overall heat transfer coefficients were assumed to be 80 for the reboilers and 100 for the condensers, all in Btu/h ft2 O F . Cooling water temperature rise of 50 OF was assumed. Results and Discussion A. SS Configuration. Tedder and Rudd (1978) suggested, without verification, that the sidestream column (SS) should be considered for use if the purity of the intermediate product stream were low. If the intermediate is present in the feed in very small amounts, it can be purged from the system in the sidestream with little loss of valuable lightest and heaviest components even if the sidestream purity is quite low. Some results for specific columns are given in Table I, where the effect of changing sidedraw purity is explored. The column had 50 trays, was fed on tray 15, and sidedraw was withdrawn from tray 23. Feed composition was 49%, 2%, and 49%. Feed flow rate was 600 mol/h. Top and bottoms product purities were maintained at 99%. Heat input is a very strong function of sidedraw purity, particularly when X S ( 2 ) increases above 60-70%. Figures 5-7 show how heat inputs change for the SS and LOF configurations as functions of intermediate feed composition for various product purities. In Figure 5, top and bottoms products are 95 mol 5%. In Figure 6, they are 99% and in Figure 7, 99.9%. Several levels of purity of the intermediate product are shown for both the SS and LOF configurations. XS(2) must be in the 50-60% range for the SS configuration to consume less energy than the conventional LOF. As top and bottoms purities are increased, the value of intermediate feed composition ZF(2) at which SS becomes better

Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 2, 1985 503

Table 11. Results for PF Configuration (High-Purity Products) concn of toluene in feed, mol % 0.04 QD (10' Btu/h) AD, ft' QB ( lo6 Btu/h) AR, ft2 NT

NF RR diameter, in. TB,

OF

TD,

OF

0.10

0.06

0.12

col. 1 col. 2 col. 1 col. 2

col. 1 col. 2 col. 1 col. 2

8.06 549 8.79 1602 23 12 1.08 85.2 348.7 236.7 0.38

8.67 597 8.74 1601 40 34/17 1.41 88 349.1 23 5.2 0.73

fixed charges on equipment (10' $/year) steam and 4.78 4.77 cooling water (10' $/year) annual cost of 5.16 5.5 svstem lo6 $/year) ' total system 10.65 cost (105 $/year)

0.16

0.20

0.30

col. 1 col. 2

col. 1 col. 2 col. 1 col. 2

8.01 544 8.7 1537 23 12 1.07 85.4 346.6 237.4 0.47

8.94 616 9.04 1655 40 34/17 1.54 89.6 349.1 235.2 0.73

8.34 558 8.94 1572 23 12 1.06 86.2 346.3 239.3 0.48

8.8 606 8.95 1638 40 34/17 1.61 89 349.1 235.2 0.73

8.51 566 9.07 1590 23 12 1.07 86.7 346 240 0.48

8.86 610 9.03 1654 40 34/17 1.69 89.2 349.1 235.2 0.73

8.48 559 8.96 2836 23 12 1.05 86.5 342.3 241.7 0.51

9.05 623 9.27 1698 40 34/17 1.05 86.5 349.1 235.2 0.74

8.47 553 8.88 2574 23 12 1.04 86 338.7 243.2 0.50

9.36 685 9.6 1763 40 34/17 2.13 91.7 349.1 235.2 0.74

8.54 9.91 543 683 8.77 10.26 2128 1879 40 23 12 34/17 1.03 2.81 86 90 330.3 349.1 247.3 235.2 0.75 0.50

4.73

4.93

4.87

4.88

4.94

4.93

3.79

5.06

3.75

5.25

3.71

5.59

5.2

5.66

5.35

5.61

5.42

5.66

4.3

5.8

4.25

5.99

4.21

6.34

10.87

10.95

11.08

10.09

10.24

10.54

Table 111. Results for SSS Configuration (High-Purity Products) ~

concn of toluene in feed, mol %

QD (10' Btu/h) AD, ft2 QB (10' Btu/h) A R , ft' NT NF R R /or L S / V S diameter, in TB,

OF

TD,OF

0.04 col. 1 col. 2

0.06 0.10 0.12 col. 1 col. 2 col. 1 col. 2 col. 1 col. 2

0.16 col. 1 col. 2

0.20 col. 1 col. 2

0.30 col. 1 col. 2

9.70 668 9.63 1763 40 13 1.70 93 349 235 0.80

10.22 704 9.90 1813 40 13 1.90 96 349 235 0.85

1.94 276 15 15 1.64 40 293 275 0.08

15.86 1093 14.37 2631 40 13 4.33 119 349 23 5 1.15

2.16 3 04 15 15 1.73 42 293 278 0.09

17.38 1197 15.03 2753 40 13 5.71 125 349 23 5 1.23

2.94 414 15 15 1.82 49.5 293 281.6 0.11

0.87 123 10 10 1.28 27 293.7 261.5 0.04

fixed charges on equipment ( l o s $/year steam and 5.26 0.35 cooling water ( i o 5 $/year) annualcost of 6.06 0.39 system ( l o 5 $/year) total system 6.45 cost (105 $/year)

1.10 157 13 13 1.37 30 2 94 263 0.06

11.59 799 1.70 1949 40 13 2.44 102 349 23 5 0.92

5.41

0.45

6.26

0.51

6.77

243 13 13 1.44 37.5 2 94 269 0.07

12.28 846 11.31 2072 40 13 2.73 105 349 23 5 0.96

1.70 241 15 15 1.54 37.5 293.8 270.5 0.08

14.07 969 12.84 2351 40 13 3.49 112 349 23 5 1.06

5.83

0.69

6.19

0.68

7.03

0.78

7.87

0.87

8.25

1.18

6.74

0.76

7.15

0.76

8.09

0.86

9.02

0.96

9.48

1.29

7.50

than LOF decreases. The vertical dashed lines in Figures 5 and 6 indicate the value of Z F ( 2 ) at which no intermediate product needs to be produced. This feed concentration decreases as top and bottoms purities increase. As the concentration of intermediate in the feed increases, the flow rate of the sidedraw increases. It can no longer be considered a "purgen, and it would carry substantial amounts of heaviest and lightest componenta with it. Rather unexpectedly, Figures 5-7 clearly show that the SS configuration becomes less attractive, compared to the LOF configuration, as the intermediate feed concentration becomes smaller. Thus it appears that the SS configuration is not suitable for systems with small amounts of the intermediate component in the feed. If Z F ( 2 ) is small, the LOF configuration requires less energy. If ZF(2) is larger, the SS configuration uses less energy but only if sidedraw

7.91

8.95

9.98

10.77

purity is low, which results in high product losses of lightest and heaviest components. The SS configuration would have a lower capital investment than the LOF since only one column is used. Therefore, the SS configuration may be the most economical in some processes where capital costs are higher than normal due to exotic materials of construction or where energy costs are extremely low. B. SSS, LOF, and PF Configurations. Detailed process data are given in Tables I1 to VI1 comparing three configurations (LOF, PF, SSS)for both low-purity (95,90, 95%) and high-purity products (99,98,99%) over a range of intermediate feed concentrations. Figures 8 and 9 show energy consumptions w. ZF(2) for the three configurations. The SS configuration is excluded from the comparison because its energy consumption is very high with these intermediate product purities.

fnd. Eng. Chem. Process Des. Dev., Vol. 24, No. 2, 1985

504

Table IV. Results for LOF Configuration (High-Purity Products) concn of toluene in feed, mol % 0.04 col. 1 col. 2

0.06 0.10 col. 1 col. 2 col. 1 col. 2

0.12 col. 1 col. 2

0.16 0.20 0.30 col. 1 col. 2 col. 1 col. 2 col. 1 col. 2 7.77 1.03 5.617 1.77 6.93 QD ( l o 6 Btu/h) 8.30 4.92 1.76 7.8 8.07 8.06 8.67 7.87 8.09 AD, ft2 572 240 534 274 535 341.21 535 345 538 555 396 426 54 2 391 1.02 5.67 8.44 8.4 7.13 QB ( l o 6Btulh) -9.07 4.96 8.49 8.19 8.6 8.39 8.31 8.8 8.24 2210 1305 2029 1500 1941 1613 1649 1510 AR, ft2 3046 908 2656 1038 2324 1281 NT 35 33 35 35 31 32 35 31. 34 31 31 35 35 35 22 10 11 23 NF 29 28 11 10 10 21 10 11 18 21 1.31 18.7 1.204 12.59 1.31 8.35 1.49 5.53 1.7 RR 1.36 ,6.75 4.53 2.03 2.38 83.1 85.3 diameter, in. 86.2 88.3 83.3 77.7 85.35 83.2 83.2 85.6 84.5 87 83.3 85.9 344.6 349.1 341.83 349.1 336.4 349.1 334.3 349.1 330.1 349.1 326.4 349.1 318.8 349.1 TB, OF TD, OF 235.2 295.5 235.16 295.1 235.2 293.1 235.2 293.7 235.2 293.7 235.2 293.1 235.2 293. I fixedcharges 0.61 0.61 0.59 0.62 0.58 0.71 0.61 0.69 0.58 0.59 0.60 0.63 0.64 0.56 o n equipment ( l o 5 $/year) steam and 3.83 2.71 3.59 3.09 3.52 3.61 3.55 3.89 3.54 4.47 3.63 4.80 3.51 4.50 cooling water (10’ $/year) annualcost of 4.54 3.32 4.28 3.69 4.12 4.21 4.16 4.5 4.13 5.09 4.21 5.44 4.07 5.13 system ( l o 5 $/year) total system 1.85 1.96 8.32 8.66 9.23 9.66 9.19 cost (105 $/year)

Table V. Results for PF Configuration (Low-Purity Products) concn of toluene in feed, mol % 0.08 QD ( lo6 Btu/h) An. ft2 Qi’(lo6 Btu/h) AR, ft2 NS,NS

NF RR diameter. in. TB, OF TD,

OF

fixed charges on equipment ( 1 0 5 $/year)steam and cooling water ( l o 5 $/year) annual cost of system ( l o 5 $/year) total system cost ( l o 5 $/year) l5

0.12

0.16

col. 2

col. 1

col. 2

col. 1

col. 2

col. 1

col. 2

col. 1 col. 2

col. 1

col. 2

7.58 51 2 8.14 2317 18 8 1.01 83 338 238 0.42

8.55 582 8.72 1527 30115 21/10 1.31 87 346 231 0.57

1.57 506 8.04 2104 18 8 0.99 82 334 240 0.41

8.26 563 8.49 1677 34/15 20110 1.41 86 346 231 0.62

7.55 499 7.94 1943 20 9 0.98 82 331 241 0.43

7.51 51 2 7.78 1362 34/20 27/15 1.31 82 346 231 0.60

1.54 493 7.86 1813 20 9 0.96 82 328 243 0.42

7.33 498 8.12 1336 34/20 25/15 1.54 83 346 237 0.60

7.52 487 7.78 1711 24 12 0.95 81 325 244 0.47

5.55 378 5.89 1031 40123 34/17 0.89 70 346 237 0.65

7.51 485 7.17 1641 23 12 0.95 81 323 246 0.45

5.6 382 5.96 1044 40124 36/17 1.07 71 346 237 0.65

3.44

4.16

3.40

4.63

3.36

4.21

3.39

4.43

3.29

3.21

3.29

3.25

3.86

5.33

3.81

5.25

3.79

4.81

3.81

5.03

3.76

3.86

3.15

3.9

9.19

9.06

8.77

8.66

’5

a 1 02

03

0.28

col. 1

I

01

0.24

0.20

04

05

06

ai

oa

09

010

ZFIZI

Figure 6. Effect of feed composition on reboiler duty (intermediate-purity products). (Purity of other product streams = 99%).

Figures 8 and 9 show that the SSS configuration consumes less energy than both LOF and PF configurations. The saving in energy is higher at low intermediate feed

7.62

I

1.65

LOF

a 01

02

03

04

05

a6

a7

08

09

oia

ZFIZI

Figure 7. Effect of feed composition on reboiler duty (high-purity products). (Purity of other product streams = 99.9%).

compositions. If energy consumption done is used to select the optimum scheme, one would choose the S S S configu-

Ind. Eng. Chem. Process Des. Dev., Vol. 24, No. 2, 1985 505 Table VI. Results for SSS Configuration (Low-Purity Products) concn of toluene in feed, mol % ~~

0.08

0.06

(lo6 Btu/h) AD, ft2 QB ( l o 6Btu/h) QD

A,, ft' NT NF RR orLS/VS diameter, in. TB,

OF

TD,"F

fixedcharges on equipment ( lo5 $/year) steam and cooling water ( l o 5 $/year) annualcostof system ( l o 5 $/year) total system cost (105 $1 Year)

0.12

0.16

0.20

col. 1 col. 2 col. 1 col. 2 col. 1 col. 2 col. 1 col. 2 10.76 8.22 9.50 1.42 505 559 140 64 I 7.79 0.37 8.20 0.14 8.86 1.34 9.94 1.48 1434 102 1551 190 1140 1362 51 198 38 7 38 40 15 40 13 15 10 7 10 13 13 13 15 15 1.25 1.28 1.37 1.76 1.00 1.49 2.33 1.70 17 81 85.5 24.6 92 33 97.8 35 346 346 290 291 294 346 289 346 254 237 257 237 264 237 269 237 0.61 0.03 0.61 0.05 0.78 0.06 0.79 0.01

11.76 801 10.76 1883 40 13 2.88 1.02 346 231 0.85

4.25

0.15

4.48

0.30

4.85

0.54

5.44

0.59

4.86

0.18

5.15

0.35

5.63

0.60

6.23

0.65

5.04

5.50

6.23

0.24

col. 1 col. 2 col. 1

1.63 217 15 15 1.87 37 288 213 0.08

12.19 850 11.31 1980 40 13 3.39 1.05 346 237 0.96

5.89

0.66

6.74

0.74

6.88

~

0.28

col. 2 col. 1 col. 2

1.77 235 15 15 2.02 38 288 275 0.08

13.39 91 2 12.21 2138 40 13 4.06 1.09 346 237 1.01

1.74 230 15 15 2.26 38 287 218 0.09

6.19

0.71

6.69

0.70

7.14

0.79

1.10

0.79

7.48

7.93

8.49

Table VII. Results for LOF Configuration (Low-Purity Products) concn of toluene in feed, mol %

0.08 QD ( lo6 Btu/h) A D , ft2 Q B ( lo6 Btu/h) A R , ft2

NT NF RR diameter, in. T B I OF T D , OF

fixed charges on equipment ( l o 5 $/year) steam and cooling water ( l o 5 $/year) annualcost of system (10' $/year) total system cost (10' $/year)

0.20

0.16

0.12

0.24

0.28

col. 1 col. 2

col. 1

col. 2

col. 1

col. 2

col. 1

col. 2

col. 1

col. 2

col. 1

col. 2

6.39 434 7.08 2162 27 20 0.76 75.2 341 237 0.46

3.39 168 3.5 600 21 6 11 77 346 292 0.31

6 .73 457 7.34 1922 21 15 0.96 77 334 231 0.39

4.69 231 4.17 835 22 7 6.08 79 346 288 0.36

6.99 416 7.56 1828 22 14 1.14 78 330 237 0.40

5.51 276 5.67 993 21 8 4.31

7.23 492 7.75 1725 21 14 1.35

7.78 516 1.77 1665 21 14 1.75 83 323 237 0.41

7.56 321 6.95 1211 22 9 2.95 85 346 292 0.41

7.35 501 7.17 1536 21 12 1.69 80 318 237 0.39

7.22 357 1.36 1289 22 10 2.29 83 346 292 0.40

2.98

1.87

3.10

2.60

3.44

2.18

3.49

2.96

5.63

81

80

346 292 0.36

326 231 0.40

6.22 308 6.34 1110 21 9 3.35 82 346 292 0.37

3.19

3.09

3.27

3.46

3.28

3.67

3.29

3.92

3.59

3.45

3.67

3.83

3.69

4.08

3.68

4.33

6.45

ration at intermediate feed compositions less than 20%. Utility costs and capital costa are given in Tables I1 to VII. The inclusion of operating costa and capital cost does not change the conclusion that was drawn from Figures 8 and 9. By examining the last row in each Table we can see that overall system cost follows the same trend as energy consumption. The SSS configuration is less expensive than the other two schemes at low intermediate feed concentrations. However, the percentage of total cost saving is less than the percentage of energy savings. This is due to the fact that higher pressure steam is used in the SSS major column, whereas lower pressure steam can be used in the first column in the LOF Configuration. The relative unimportance of capital costs is illustrated in Table VIII. Despite the large savings in capital cost shown for SSS over LOF, the overall cost is dominated by energy cost. Conclusions Sidestream columns with strippers should be considered when designing a distillation column configuration for

7.04

7.51

7.78

8.00

Table VIII. Cost Comparison between SSS and LOF Configurations ZF(2) = 0.08, ZF(2) = 0.04, high-purity % diff between low-purity products producta SSS and LOF 18.4 33.6 energy consumption energy cost 1.4 16.6 capital cost 8.25 56.0 2.36 22.0 total cost

separating ternary mixtures with small concentrations of the intermediate component in the feed. The range of intermediate feed compositions for which the SSS configuration is optimum increases as the purities of the products decrease. It is particularly sensitive to the purity of the intermediate product.

Future Work The next step in the reaearch is to explore the dynamics and controlability of the SSS configuration for disturbances in feed composition and rate. A dynamic comparison

Ind. Eng. Chem. Process Des. Dev. 1985, 2 4 , 506-507

506

d 6(

04

08

12

16

20

inlermediele leed concentration ZF

24

I (mole

28

30

% 01 IolUeneI

Figure 8. Comparison of PF, LOF, and SSS configurations (lowpurity products). (Product purities 95%, 90%, 95%). *Ot

NTS = total number of trays in sidestripper F = feed rate, mol/h LS = sidedraw rate, mol/h VS = sidestripper net vapor feed to the main column, mol/h SS = sidestripper bottoms flow rate, mol/h ZFG) = jth component feed composition X S G ) = middle products composition in SS or SSS configuration X L G ) = sidestream stripper liquid feed composition X D l G ) = distillate composition from first column of LOF scheme XD2G) = distillate composition from second column of LOF scheme X S l G ) = bottoms composition from first column of LOF scheme XB2G) = bottoms composition from second column of LOF scheme T B = bottoms product temperature, O F TD = distillate temperature, OF QB = reboiler duty, Btu/h QD = condenser duty, Btu/h QBs = sidestripper reboiler duty, Btu/h TBp = bubble point temperature Tk = temperature at the kth iteration RR = reflux ratio DIAM = column diameter, in. A R = reboiler area, ft2 A D = condenser area, ft2 wt = weighting factor for bubble-point method convergence

Literature Cited

65

010

015

020

Intermedial8 leed concenlmon. Z F

025 ~

030

lmole R loluenel

Figure 9. Comparison of PF, LOF, and SSS configurations (intermediate-purity products). (Product purities 99%, 98%, and 99%).

of the LOF and SSS configurations will be reported in a later paper. Nomenclature NT = total number of trays NF = feed tray N S = tray of sidestream drawoff

Bumingham, D. W.; Otto, F. C. Hydrocarbon Process. 1967, 46. 10. Dryden, C.: Furlow, R. “Chemical Engineering Costs”: The Ohio State University: Columbus, OH, 1966. Doukas, N.; Luyben, W. L. Ind. Eng. Chem. Process Des. D e v . 1976, 17, 272. Elaahl, A.; Luyben, W. L. Ind. Eng. Chem. Rocess Des. D e v . 1983, 22, 80. Holland, C. D. ”Fundamentals of Mukicomponent Distlllatlon”; McGraw-Hill: New York, 1981. Petlyuk, F. B.; Platonov, V. M.; Slavinski, D. M. Int. Cbem. Eng. 1965. 5 , 3 . Seader, J. D. “Mathematical Modeling for Process Design”; AIChE Short Course, Houston, 1981. Tedder, D. W.; Rudd, D. F. AICbE J . 1976, 2 4 , 303. Wang, J. C.; Henke, G. E. Hydrocarbon Process. 1967, 45. 8 .

Department of Chemical Engineering Lehigh University Bethlehem, Pennsylvania 18015

Imad M. Alatiqi William L. Luyben*

Received for review May 23, 1983 Revised manuscript received February 21, 1984 Accepted June 11, 1984

Effect of Raw Oil Shale Grade on the Kinetics of Oxidation of Carbonaceous Residue in Retorted Shale The effect of the grade of raw oil shale on the intrinsic kinetics of oxidation of carbonaceous residue formed during retorting was investigated. The resutts indicate that, within the experlmental uncertainty, the intrinsic reactivity of the carbonaceous residue of Colorado oil shale is independent of orlglnal grade.

In a previous article (Sohn and Kim, 1980), the authors presented the result of an investigation in which the intrinsic kinetics of oxidation of carbonaceous residue in retorted shale were determined. In this communication we discuss the effect of raw oil shale grade on the kinetics of this reaction. All of the experiments reported in the authors’ previous article (Sohn and Kim, 1980) were carried out with samples prepared from a 39.4 gal/short ton oil shale from the Anvil 0196-4305/85/1124-0506$01.50/0

Points Mine in Colorado. In order to determine whether the grade of oil shale has any effect on the kinetics of char oxidation, retorted shales were prepared from oil shale blocks of 57 and 21 gal/short ton from the same source. Samples containing approximately 8% carbonaceous residue were obtained from the former and 3% carbonaceous residue from the latter. The char contents were measured by oxidation at low temperature (