JIANG Zhankun (姜占坤) and BAI Peng (白鵬),*

12 School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, China Key Laboratory of Systems Bioengineering, Ministry of Education, Tianjin University, Tianjin 300072, China

1 INTRODUCTION

Batch distillation is widely used in the production of small volume, high value added products in pharmaceutical and fine-chemical industries because of its high flexibility [1-3], low equipment cost, the capability of separating multi-component mixtures with only a single column and ease of frequent change of feed composition in production.

The cyclic total reflux (CTR) operation mode,with which the column presents the highest separation efficiency most of the time during the operation and simple control, is one of the most promising modes for batch distillation. In the simplest way of the total reflux operation proposed by Treybal [4], Bortolini and Guarise [5], the binary feed mixture is split between the reflux drum and the reboiler. The holdups of the reflux drum and the reboiler are correctly chosen in terms of the feed concentration. The column runs under total reflux all the time until the heavy components of high purity are obtained. Baroloet al. [6, 7]applied this total reflux mode to a middle vessel column. Gruetzmannet al. [8] studied the operation in the middle vessel column with cyclic operation. Leipoldet al. [9, 10] described an evolutionary approach for multi-objective dynamic optimization applied to the middle vessel batch distillation. Cuiet al. [11],Phimister and Seider [12] studied the batch extractive distillation in a column with a middle vessel.

Further improvements were made in recent decades by repeating total reflux (TR) and total withdrawal (TW) for several cycles [13], which was named the cyclic total reflux batch distillation (CTRBD).This mode has three main periods for each cycle: filling up, total reflux and dumping, as shown in Fig. 1.The first and third periods are TW operation, while the second period is TR operation. The cyclic policy is easy to operate and yields satisfactory separation [14].Those difficult separations with low amount of light component in a traditional regular column are most likely to give large reductions in operating time when changing from a constant or optimal reflux ratio policy to the cyclic policy. Baiet al. [15] proposed a two-reflux-drum mode of CTRBD operation, in which the column has two reflux drums working in turn at

Figure 1 Three periods in cyclic total reflux operation D—distillate rate; L—liquid flow rate; V—vapor flow rate

* To whom correspondence should be addressed. E-mail: bp2008@eyou.com the top. The operation is characterized by repeating two operation steps as a cycle: filling the drum (TW operation) and running at TR (the second step). This new mode is time-saving according to the experiments and simulations. Baiet al. [16] put forward a dual temperature control method on CTR operation without reflux drum, showing experimentally that the control method is feasible and effective.

This paper presents the computational results and experimental data of the “overhead concentration platform” (OCP) in TW operation, which is a period of time with high overhead concentrations when the operation is changed from total reflux to total withdrawal. The factors affecting the OCP, such as the theoretical stage number, feed concentration and different feed mixtures, are investigated.

2 MATHEMATICAL MODEL

For the simulation of the OCP, the dynamic mathematical model proposed previously [17] is used.This rigorous model considers the tray hydraulics during the filling drum period and has been validated by experiments.

In the simulation, the operation is switched from TR to TW after equilibrium is achieved. In the TR operation, the column concentration profile is built and the overhead concentration remains at high values.Because the theoretical stage number in the column is larger than the minimum theoretical stage number,after sufficient TR operation, the upper part of the column is filled with high-concentration liquid. Besides, because of the “fly wheel effect”, the overhead concentration will remain at high values for a short period of time when it is switched from TR to TW operation.

When the TW operation begins, the overhead concentration is higher than the desired purity of product. Then the overhead concentration decreases to the desired product purity, and the period is defined as the OCP time.

3 SIMULATION RESULTS

Table 1 shows the specifications for the computation. For comparing the simulation with experiments,all model parameters used in the simulation are determined separately in laboratory experiments.

Table 1 Specifications for computation

3.1 Influence of composition

Different mixtures may have different relative volatility. Fig. 2 shows the simulation results for different relative volatilities while the feed concentration is 0.5. When the relative volatilityαis larger, the OCP time is longer, but the difference in OCP time is relatively small.

Figure 2 Overhead concentration vs. time for different relative volatilitiesα: ○ 1.72; ■ 2.1

3.2 Influence of theoretical stage number

The influence of theoretical stage number (N) on the platform is investigated for relative volatility of 2.1 and feeding concentration of 0.5. Fig. 3 shows that the OCP time increases remarkably as the stage number increases.

Figure 3 Overhead concentration platform with different stage numbersN: ○ 15; ■ 20

3.3 Influence of feed concentration

As shown in Fig. 4, the OCP time increases with the feed concentration (FC, molar fraction). After the OCP period, the overhead concentration decreases with time in different modes for different feed concentrations. The figure also indicates that the feed concentration has relatively small influence on the OCP width.

Figure 4 OCP at different feed concentrations of ethanol/1-propanolFC: ▲ 0.0765; ○ 0.25; △ 0.5; ● 0.75

4 EXPERIMENTAL

4.1 Experimental apparatus and procedure

The experimental apparatus for measurement of OCP is shown in Fig. 5. It consists of a still (reboiler)of 2 L in capacity, a condenser, and a packed column of 50 mm inner diameter. The still pot is connected to the glass column through a taper joint. Heat insulation material is wrapped around the glass column to minimize the thermal losses. The column is packed with two-layered stainless steel Dixon Rings of 5 mm in diameter and 5 mm in length. The two-layered packing is specially used to increase the tray holdup. The height of the packed section is 1000 mm. Measured by the mixture ethanol/isopropanol, the number of theoretical stages is 20. At the end of the heating-up under infinite reflux ratio (R), the liquid hold-up is measured for 5 times. The whole hold-up in the column is 211.8 ml on average, which means the tray holdup is 10.6 ml.By opening and closing valve 10, the operation state is switched repeatedly between TR and TW operations.

Figure 5 Scheme of experimental apparatus1—heater; 2—round bottom flask; 3, 6—temperature sensor;4, 7, 9—sampling point; 5—packed section; 8—condenser;10—valve; 11—product receiver; 12—U-tube manometer

The operation started under TR with a constant pressure drop. After the temperatures in the still pot and condenser became invariable, the samples were collected from sample points 7 and 4 every 30 min.When the sample concentration remained constant, the column reached the equilibrium state. Then the operation was switched to TW with the distillate collected in the product receiver. The distillate sample was collected every 30 s from the sample point 9. The sampling was stopped when the overhead temperature was obviously higher than the equilibrium temperature.

The experiments were carried out with three mixtures, methanol/ethanol (α=1.72), ethanol/1-propanol(α=2.10) and methanol/water. For the ethanol/1-propanol mixture, the experimental OCP time was compared at different feed concentrations, and the influence of the number of theoretical stages was investigated. The liquid-sample compositions were determined using gas chromatography.

4.2 Experimental results and discussion

4.2.1Different feeding mixtures

Three mixtures, methanol/ethanol, ethanol/1-propanol, and methanol/water, are used to examine the influence of relative volatility on OCP. The methanol/water mixture is non-ideal solution and its relative volatility varies with concentration. Fig. 6 shows the change of the overhead concentration with time for the mixtures. The OCP time is 7.5 min for the methanol/ethanol mixture, while it is 10 min for the ethanol/1-propanol mixture. Thus larger relative volatility leads to longer OCP time, consistent with the simulation results. The general behavior of the process is predicted with sufficient accuracy. For the non-ideal mixture methanol/water, there also exists an OCP since the two reasons, the high concentration liquid in the upper part of the packed column and the “fly wheel effect”, still hold true. Its OCP time is the longest among the mixtures, because its average relative volatility is the largest.

Figure 6 Overhead concentration with time for the mixtures■ ethanol/1-propanol; ○ methanol/ethanol; △ methanol/water

Figure 7 shows the temperature in the methanol/ethanol separation. The top temperature is unchanged in the first 8.5 min. In the next 11.5 mins, it rises with an increasing speed. Thus the overhead temperature forms a similar platform as the OCP. Because the bottom concentration varies little, the bottom temperature is almost unchanged in 20 min.

Figure 7 Overhead concentration, top and bottom temperatures with time for the mixture methanol/ethanol△ top temperature; ○ bottom temperature; ■ concentration

4.2.2Influence of feed concentration

Figure 8 Overhead concentration with time at different feed concentrationsFC: ▲ 0.0765; ○ 0.25; △ 0.5; ● 0.75

Figure 8 displays the OCP at the feed concentrations of 0.076, 0.25, 0.5, and 0.75 for the mixture ethanol/1-propanol. For the feed concentration of 0.076, the OCP time is short and the overhead concentration drops dramatically to its lowest value as the TW operation lasts a long time. At higher feed concentration, the OCP time is longer, which agrees well with the simulation results. It also indicates that the decrease pattern of the temperature changes depends on the feed concentration.

4.2.3Influence of theoretical stage number

Figure 9 shows the overhead concentrations for the theoretical stage numbers 20 and 15 with the mixture ethanol/1-propanol. The stage number is an important factor that influences the OCP time, which agrees well with the simulation results.

Figure 9 Overhead concentrations with time for different stage numbersN: ■ 20; △ 15

5 POTENTIAL APPLICATION OF OCP

The OCP phenomenon is useful for the selection of suitable reflux drum volume in a given CTR operation, in which the filling drum period is the TW operation. Because of the OCP phenomenon, the filling liquid in the drum is of high concentration. The drum volume can be determined accordingly.

Figure 10 Two characteristic periods of no-reflux-drum cyclic total reflux operation D—distillate rate; L—liquid flow rate; V—vapor flow rate

The OCP phenomenon can be used to evaluate the feasibility of the cyclic TR operation without reflux drum. The no-reflux-drum CTR operation is shown in Fig. 10. In the operation, the column is operated at TR until the overhead concentration reaches a specified value. Then the second step, the TW operation, is conducted. Because of the OCP, on-purity product can be drawn off from the top of the column in the TW operation. When the overhead concentration is lower than the specified value, the operation is switched to TR operation. This two-step-procedure is repeated in the operation until either the light component in the product receiver or the heavy component in the still meets the requirement. The no-reflux-drum CTR operation has two obvious characteristics: (1) enriching the light component with the highest separation efficiency (TR) and (2) drawing off the top products at the highest speed (TW).

6 CONCLUSIONS

An interesting phenomenon, the overhead concentration platform, in the TW operation is studied.Both simulations and experiments confirm the overhead concentration platform. The results show that the number of theoretical stages influences the platform time more significantly than the feed concentration and composition.

1 Zhao, S.Y., Huang, J.Z., Wang, L.E., Huang, G.Q., “Coupled reaction/distillation process for hydrolysis of methyl acetate”, Chin. J.Chem. Eng., 18, 755-760 (2010).

2 Li, C.L., Fang, J., Wang, H.H., Sun, L.J., “A quasi-steady-state model for numerical simulation of batch extractive distillation”,Chin. J. Chem. Eng., 18, 43-47 (2010).

3 Jahanmiri, A.H., Khazraee, S.M., “Composition estimation of reactive batch distillation by using adaptive neuro-fuzzy inference system”, Chin. J. Chem. Eng., 18, 703-710 (2010).

4 Treybal, R.E., “A simple method for batch distillation”, Chem. Eng.,10, 95-98 (1970).

5 Bortolini, P., Guarise, G., “A new method of batch distillation”,Quad. Ing. Chim. Ital, 6, 1-9 (1970).

6 Barolo, M., Botteon, F., “Simple method of obtaining pure products by batch distillation”, AIChE J., 43, 2601-2604 (1997).

7 Barolo, M., Guarise, G.B., Rienzi, S.A., Trotta, A., Macchietto, S.,“Running batch distillation in a column with a middle vessel”, Industrial & Engineering Chemistry Research, 35, 4612-4618 (1996).

8 Gruetzmann, S., Fieg, G., Kapala, T., “Theoretical analysis and operating behaviour of a middle vessel batch distillation with cyclic operation”, Chem. Eng. Process., 45, 46-54 (2006).

9 Leipold, M., Gruetzmann, S., Fieg, G., “An evolutionary approach for multi-objective dynamic optimization applied to middle vessel batch distillation”, Computers & Chemical Engineering, 33, 857-870(2009).

10 Leipold, M., Gruetzmann, S., Fieg, G., “Model based optimisation of the design of a middle vessel batch distillation by using an evolutionary algorithm”, Forschung Im Ingenieurwesen-Engineering Research, 73, 161-172 (2009).

11 Cui, X.B., Yang, Z.C., Zhai, Y.R., Pan, Y.J., “Batch extractive distillation in a column with a middle vessel”, Chin. J. Chem. Eng, 10,529-534 (2002).

12 Phimister, J.R., Seider, W.D., “Semicontinuous, middle-vessel, extractive distillation”, Comput. Chem. Eng., 24, 879-885 (2000).

13 S?rensen, E., Skogestad, S., “Optimal operating policies of batch distillation”, Proceedings and Symposium, PSE'94. Kyongju, Korea,449-453 (1994).

14 S?rensen, E., “A cyclic operating policy for batch distillation-theory and practice”, Comput. Chem. Eng., 23, 533-542 (1999).

15 Bai, P., Hua, C., Li, X.G., Yu, K.T., “Cyclic total reflux batch distillation with two reflux drums”, Chem. Eng. Sci., 60, 5845-5851 (2005).

16 Bai, P., Li, X.F., Sheng, M., Song, S., Zhao, G.P., Li, X.G., “Study of dual temperature control method on cyclic total reflux batch distillation”, Chem. Eng. Process., 46, 769-772 (2007).

17 Bai, P., Song, S.A., Sheng, M., Li, X.F., “A dynamic modeling for cyclic total reflux batch distillation”, Chin. J. Chem. Eng., 18,554-561 (2010).