Total Reflux Operation of Multivessel Batch Distillation

I N T R O D U C T I O N Although batch distillation generally is less energy efficient than continuous distillation, it has leceived increased attention in the last few years because of its simplicity of operation, flexibility and lower capital cost. For many years academic research on batch distillation was focused primarily on optimizing the reflux policy. However, in most cases the difference to the simple-minded constant reflux policy usually is small. More recently, one has started re-examining the operation of batch distillation as a whole. The total reflux operation of a conventional batch distillation column was suggested independently by Bortolini and Guarise (1971) and Treybal (1970). A generalization is a cyclic operation where the operation is switched between total reflux operation and dumping the product (i.e., the condenser holdup is introduced as an additional degree of freedom) may be better (Scrensen and Skogestad, 1994). The simplest operation strategy is with only one cycle, that is, the column is operated under total reflux and the final products are collected in the condenser drum and in the reboiler. Another alternative is to "invert" the column by charging the feed to the top and removing the heavy product in the bottom (Robinson and Gilliland, 1950; SCrensen and Skogestad, 199.5). It has also been suggested to use a middle vessel where the feed is charged to the middle of the colunm (Bortolini and Guarise, 1970). Hasebe et al. (1995) extended this idea and proposed a multivessel column with total reflux operation where one can separate more than two components. They denoted this a "multi-effect batch distillation systenf'. Q c ~ ~ All these policies may be realized in the multivessel batch distillation column shown ill Figure 1 A l ~ with both holdups and product flows as degrees of v i . 1 freedom. With Nc vessels along the cohnnn and with given pressure and heat input,, this column has 2N~ 1 degrees of freedom for optimization; namely the Nc 1 holdups (e.g., controlled by the N,. 1 reflux streams) and the N~ product rates. D 2 The simplest operatiort form of the proposed mulv ~2 tivessel column, which is the focus of this paper. is the total reflux operation suggested by Hasebe et.al.(1995) where the Nc product rates are set to zero (Di --0). There are at least two advantages with this multivessel column compared to conventional batch ~ 3 products are taken ow~r the top, distillation where the v _ _ L 3 one at a time. First, the operation is simpler since no product, change-overs are required during operation. Second, tile energy requirement may be nmch less due to the multi-effect nature of the operation (ttasebe et a1.,1992), where the heat required for tho separation is supplied only to the reboiler and cooling is done only at tile top. In fact, Hasebe et al. (1995) 9 4 show that for some separations with many components the energy requirement be similar to that may for continuous distillation using Nc 1 co]mnns. Figure 1: Multivesse] batch distillation column Hasebe et al. (1995) propose to "control" the total reflux multivessel batch distillation colmnn by calculating in advance the final holdup in each vessel and then using a level control system to keep the 1 Centre for Process Systems Engineering, ImperiM College, London, SW7 2BY 2 Author to whom correspondence should be addressed, fax: +47 7359 4080; E-math [email protected] Sl041 S1042 European Symposium on Computer Aided Process Engineering--6. Part B holdup in each vessel constant. For cases where the feed composition is not known exactly they propose to, after a certain time, adjust the holdup in each vessel based on composition measurements. Their scheme, involving the optimization of the vessel holdups and their adjustment based on composition measurement in these vessels, is rather complicated to implement and requires an advanced control structure to implement the control law. We propose a feedback control structure based on N~ 1 temperature controllers (see Fig. 3). The idea is to adjust the reflux flow out of each of the upper Nc 1 vessels by controlling the temperature at some location in the column section below. There is no explicit level control, rather the holdup in each vessel is adjusted indirectly by varying the reflux flow to meet the temperature specifications. The remainder of this paper is divided into 6 parts. First we present the principle of operation and a simulation example to show the feasibility of the proposed process. The dynamic models are implemented in the SPEEDUP environment (Speedup, 1993). In the second part the proposed implementation of the multivessel colunm and dynamic simulation results of its operation are given and in part 3 we outline the experimental setup. The procedure to operate the multivessel batch column and experimental results are presented in section 4 and compared to simulations. Finally the discussion and conclusions are given. T O T A L R E F L U X O P E R A T I O N In this section we follow Hasebe et al. (1995) and present simulations which demonstrate the feasibility of the nmltivessel batch distillation under total reflux. The holdup of each vessel is calculated in advance by taking into account the amount of fee& feed composition and product specifications. After feeding the predescribed amount of raw material to the vessels, total reflux operation with constant vessel holdup is carried out until the compositions in all vessels satisfy their specifications. Data for the column and feed mixture are given in Table 1. Table 1 : Summary of column data and initial conditions Number of components N~ = 4 Relative volatility ai = [10.2, 4.5, 2.3, 1]* Total number of stages Ntot = 33 Number of sections N~ = 3 Number of stages per section Art = 11 Vessel holdup M,n = 2.5 kmol Tray holdup Mt = 0.01 kmol Total initial charge Mtot = 10.33 kmol Boilup ratio Mtot/V = 1.03 hr Reflux flow L = 10 kmol/hr Vapor flow V = 10 kmol/hr * The numerical value of ratios of the relative volatilities are chosen to be close to the experimental system (methanol, ethanol, n-propanol, n-butanol) in the pilot plant. Constant molar flows are assumed. Typical simulated composition profiles as a function of time are shown in Figure 2 for a 4-component mixture with an initial feed composition of zr, t = [0.25, 0.25, 0.25, 0.25] (1) Table 2: Steady state composition for initial feed composition ZF,1;Constant vessel holdups Mi[kmol] I l[ Vessel 1 ] Vessel 2 I Vessel 3 I Vessel 4 I Mi 2.5 2.5 2.5 2.5 xl 0.993 0.017 0.0 0.0 xz 0.007 0.959 0.025 0.0 x3 0.0 0.024 0.963 0.004 x4 0.0 0.0 0.012 0.996 Table 3: Steady state composition for initial feed composition zF,2; Constant vessel holdups Mi[kmol] II Vessel 1 I Vessel 2 Mi 2.5 2.5 xl 0.999 0.203 x2 0.001 0.404 x3 0.0 0.393 X 4 0 .0 0.0 Vessel 3 I Vessel 4 I 2.5 2.5 0.0 0.0 0.001 0.0 O.999 0.180 0.0 0.820 As time goes to infinity the steady state compositions presented in Table 2 are achieved. However, the operation policy of keeping the holdup of the vessels constant may be difficult to achieve in practice and also is very sensitive to errors in the assumed feed composition. The last problem is illustrated by considering a case where the actual feed composition is zF,~ = [0.30, 0.10,0.40,0.20] (2) but the holdup of each vessel is equal to the example with feed composition ZF,1 in Eq. 1. This results ill large changes in the final vessel compositions as seen from Table 3. For example, the purity in vessel 2 is reduced from x2 = 0.959 to x2 = 0.404. To compensate for these feed variations Hasebe et al. (1995) propose a rather complicated algorithm for adjusting the holdup based on measuring the composition in the vessels. We propose a much simpler scheme which is discussed in the next section. A N E W F E E D B A C K C O N T R O L S T R U C T U R E A flowsheet of our proposed control structure is shown in Figure 3. The separation of a mixture containing European Symposium on Computer Aided Process Engineering--6. Part B S1043 Nc components require N¢ vessels and N~ 1 temperature controllers. The temperature controller ( e.g.: Tc= ) adjust the reflux flow ( e.g.: L2 ) out of the vessel ( e.g,: M2 ) above that column section, This enables an indirect control of the holdups in the vessels. Note that there is no level controller or level measurement, although some minimum and maximmn level sensors may be needed for safety reasons.