Dynamic response of a binary distillation column

Dynamic response of a binary distillation column

Chemical Engineering Science, 1974, Vol. 29, pp. 2253-2256. Pergamon Press. Printed in Great Britain Dynamic response of a binary distillation...

253KB Sizes 22 Downloads 210 Views

Chemical

Engineering

Science,

1974, Vol. 29, pp. 2253-2256.

Pergamon

Press.

Printed

in Great Britain

Dynamic response of a binary distillation column (Received

23 March

1973; accepted

The dynamic response characteristics of a binary distillation column have been discussed extensively in the recent literature. Archer and Rothfus [I], Williams [2], Holland [3] and recently Fell[4] have presented detailed literature reviews. It is apparent from these reviews that the major effort has been devoted to a theoretical analysis of the binary distillation process and that, in general, the proposed models have been linear in nature and subject to a large number of simplifying assumptions. Other authors[5], Armstrong and Wood[6], Barber et al. [7], Osborne et al. [8], Huckaba et al. [9, 101, Luyben and Gerster[ll], Burman and Maddox[l2] have attempted to provide experimental verification of their proposed models and have claimed reasonably good agreement between the predicted and measured response. Unfortunately, in

3 May

1974)

most cases, the experimental results were not presented. Consequently, it was impossible to assess the quality of this agreement in a quantitative fashion. In addition, the majority of these investigations considered only the effect of small load disturbances which did not clearly demonstrate the nonlinear response characteristics of the system. A further limitation was imposed by the fact that composition measurements, performed during the transient period of operation, were restricted to a very few locations in the column and/or required the removal of appreciable quantities of liquid sample from the column. This latter practice was shown to generate serious load disturbances which markedly influenced the response of the column[l3]. This paper presents the theoretical and experimental

H

nr-

Composition recorder

O\;erheod

I S&am p~o~t$c~s condensate Fig. 1. Hardware 2253

configuration.

Shorter

2254

Solution

Solution \7 L

from

Communications

to plote

plate

Capocltonce meosurlng heod

Temperature compensator odlustments

Fig.

2. Continuous

Cawcitonce

sampling and system.

capacity,

recorder

recording

transient response of a binary distillation column designed to separate methanol and water to produce a final product of approximately 98 wt.% of methanol. The experimental apparatus, illustrated schematically on Fig. 1, consisted essentially of an eight tray, 0.23 m dia glass distillation column insulated with two layers of lucite tubing and provided with a total condenser and a basket type reboiler. For this study, the reflux flow rate was maintained constant with the aid of a simple flow control loop while the remaining variables were controlled in the manner indicated on Fig. 1. The trays were of the bubblecap type and were provided with suitable ports for sampling and temperature measurement[l3]. The composition of the bottom product, the overhead condenser product and the liquid on each tray was continuously monitored with the aid of a specially designed multipoint capacitance recorder. Liquid samples Table 1. Source Reference figure number 4 5

Type

of step

Feed flow rate step change to Feed composition step change to

from each tray were circulated continuously through individual, thermally compensated capacitance cells. A schematic illustration of the complete monitoring arrangement is provided in Fig. 2. In view of the large variations in the capacitance of methanol-water solutions[l3], accurate measurement of changes in composition were achieved with relative ease (? 0.0025 st fraction). The performance of this monitoring system is clearly demonstrated by the excellent reproductibility of the results presented on Fig. 3. In order to maintain a check on the capacitance system, two thermocouples were located on each tray. One thermocouple was placed in the center and one on the edge of the tray. These temperatures indicated a composition variation of less than * 0.005 wt fraction. This is within the accuracy of the composition measurement, hence the conclusion that there is no concentration gradient on the tray. In addition to the measurement of the various liquid compositions, a continuous record of all pertinent flow rates and inventories was maintained. It was, therefore, possible to perform a complete steady state and transient analysis of the system under a wide variety of operating conditions. In order to simplify the development of a mathematical model, the distillation column was considered as a combination of interacting elements, associated with the trays, the reboiler and the condenser, which could be adequately described in terms of appropriate lumped parameters. The detailed development of the mathematical model can be found in an earlier work[l3]. This reference also contains the details of the computer simulation of the mathematical model. To simulate 100 min of real time on the IBM 7044 took 5 min. This time was reduced to 2.5 min by running on the IBM 360-65. For time steps of one min or greater the integration procedure would not converge. By using a time step of l/8 min, the integration procedure took about ten iterations for early time steps and reduced to 3 iterations at around l/2 the response period and finally to one or two for the remainder. To check the reproductibility of the results, the simulation was run with an integration time interval of l/l6 min. The same results as for a l/8 min integration interval were produced by the program. In order to test the validity of the open loop dynamic model, the predicted response of the distillation column was compared with the observed response under a variety of operating conditions. More specifically, the transient behaviour of the liquid composition on each tray was determined following a significant step change in the feed flow rate, and the feed composition. The magnitude of each variable and the associated step change is presented on Table 1 while the corresponding experimental am

and magnitude

of load disturbances

Flow rates (Kg/set Feed Steam 15.38 8.06 18.94

x 10’) Reflux

Feed composition (Wt. fraction)

11.24

12.5

0.5151

14.82

17.2

0.4584 0.5599

2255

r

n RUN RR-I 0 RUN RR-3 Product 2 Tmy 8 A Tmy 7 n Tray6 A Tray 5

a Tray 4 fi Tray3 (feed) 6 Troy 2

&Troy

I

Note: RR-3 is platted only where values differ from RR-I by more than 0,005 weight fraction. gn(LD b~llrrlllll O0 ‘YO

20

30

n

a

n

0

4

40

50

60

70

130

Time,

AReboIler 90,

100

min

Fig. 3. Reproducibility open loop, transient response step increase, in reflux flow rate.

Product Tray 8 Tray 7

Tray 6

IO

20

30

40

50

60

70

80

90

tm

ReboLer Time,

min

Fig. 4. Open loop transient response step decrease in feed flow rate.

2256

Shorter

Communications

Product Jr”,;87 Troy 6 Tray 5

Time, Fig. 5. Open loop transient

response ‘step increase

predicted transient response curves are illustrated on Figs. 4 and 5. In spite of the rather large load disturbances imposed on the system, it is apparent that the agreement between the predicted and measured response is quite satisfactory. The average overall error is about one percent. The discrepancies which are evident in some cases probably result from a neglect of the transport lag associated with the flow of liquid through the downcomers and the inadequate representation of the tray efficiencies. Faculty of Engineering University of Calgary

min

W. Y. SVRCEK

in feed composition.

Applications in Multicomponent Distillation, Chapt. 4, Prentice-Hall, New Jersey 1966. [4] Fell C. J. D., Distillation. . The State of the Art. Jf Inst. Eng., 1971 3-6. [5] Rademaker 0. and Rijnsdorp J. E., Proc. 5th World Pelr. Congr., Paper 5, Sec. VII May 1959. [6] Armstrong W. D. and Wood R. M., Chem. Engng Sci. 1960 12 272; Trans. Inst. Chem. Engr 1961 39 65. [7] Barber M. F., Edwards L. L. Jr., Harper W. T. Jr., Witte M. D. and Gerster J. A., Chem. Engng Progr. Symp. Ser., 1961 36, 57, 148-159. [8] Osborne J. R., Reynolds D., West J. B., Maddox R. N., A.I.Ch.E.-Inst. Chem. Engr.-1. Adv. Sep. Tech.-Paper 1.8 London, June 13-17,1965,67-72. [9] Huckaba C. E., May F. P. and Franke F. R., Chem.

Calgary, Alberta U.S.A.

Engng Symp. Ser. 1963 46, 59, 38-47. Faculty of Engineering Science University of W. Ontario London, Ontario, Canada.

R. A. RITTER

REFERENCES

R. R., Chem. Engng Progr. Symp. Ser., 1961 No. 36, 57 2-19. [2] Williams T. J., Chem. Engng Symp. Ser. 1%3 46, 59, [I] Archer

1-8. [31 Holland

D. H. and Rothfus

C.

D.,

Unsteady

State

Processes

with

[lo]

Huckaba C. E., Franke F. R., May R. P., Fairchild B. T. and Distefano G. P., Chem. Engng Progr. Symp. Ser. 1965 55, 61, 126135. [ll] Luyben W. L. and Gerster J. A., Ind. Engng Chem. Proc. Des. Dev. 1964 3 374. [I21 Burman L. K. and Maddox R. N., Dynamic Control of Distillation Columns, I.E.C. Proc. Des. and Dev. 1969 8, 433. [13] Svrcek W. Y., Dynamic Response of a Binary Distillation Column, Ph.D. Thesis, University of Alberta, Edmonton, 1967.