SYNTHESIS/DESIGN/OPTIMIZATION OF ETHANOL

DEHYDRATION USING BENZENE

Víctor Beckley, Carlos A. Parodi, Enrique A. Campanella ∗

���������Heterogeneous azeotropic distillation is a process of industrial interest for the separation of binaryazeotropic mixtures into their constituent pure components. This technique works by adding an entrainer thatcauses liquid-liquid-phase separation over a broad range of compositions in the ternary phase diagram. Thisliquid-liquid-phase split provides a cheap and efficient method for moving across distillation boundariescaused by the presence of azeotropes in the mixture. Most of the available methods for synthesizing aheterogeneous distillation system use heuristics to determine the system structure. This work describes asystematic procedure for generating process alternatives for multicomponent heterogeneous distillation.Alternatives based on distillation and decanting are expanded by mixing and by recycling to obtain productswith high purity. The separation of ethanol from water using benzene as the entrainer is used to illustrate theprocedure. Comparisons are made with competing design from the literature. Phase equilibrium and processcalculation were done with DISTIL and HYSYS. The thermodynamic model used includes ideal gas phaseand pure liquid state as reference for the chemical potentials. UNIQUAC and NRTL models were used andcompared before doing process calculation. Prior to comparisons the different design are optimized to drawconclusions on the relative merits of the different sequences.

�� ����� Ethanol Dehydration, Heterogeneous Azeotropic Distillation, Sequence.

1.� ������������

Heterogeneous azeotropic distillation is a process of industrial interest for the separation of binary azeotropic

mixtures into their constituent pure components. This techniques works by adding an entrainer that causes

liquid-liquid phase separation over a broad range of compositions in the ternary phase diagram. Heterogeneous

azeotropic distillation has been broadly discussed in the literature (Pham and Doherty, 1990a, b, c; Widagdo and

Seider, 1996). In particular, Feng et al. (2000) and Tao et al. (2003) developed different approaches to process

synthesis of heterogeneous azeotropic column sequences. Feng et al. (2000) developed an approach based on

graph theory and combinatorial techniques. Tao et al. (2004) proposed an approach that combines several

procedures including simple rules to treat recycles. Furthermore, in the literature there are available several

patents involving alternative configurations for ethanol dehydration sequences. Recently a surprising feature of

azeotropic distillation columns, multiple steady states, was discovered. It refers to output multiplicities, that is,

that a column of a given design exhibits different columns profiles at steady state for the set of inputs and the

same values of the operating parameters. The existence of multiple steady states in heterogeneous azeotropic

distillation is well known and the study of output multiplicity in distillation has been carefully done (Bekiaris et

al., 1996; Bekiaris et al., 2000).

In this paper we apply first a synthesis procedure to generate separation sequences for ethanol dehydration

using benzene, and then, a selection of sequences are optimized before comparing them.

∗ To whom all correspondence should be addressed.Address: INTEC (UNL - CONICET) - Güemes 3450 - S3000GLN Santa Fe - ArgentinaE-mail: Tquique@ceride.gov.ar

������ Composition diagram for ethanol-water-benzene.

2.� ����������������

Before doing synthesis, design and optimization we have modeled the phase equilibrium of the ethanol -

water - benzene system in two steps: first we estimates binary parameters using all information we have at our

disposal, and second we design a distillation column to see if the parameters were adequate. We modeled the

liquid phase with UNIQUAC and NRTL. Our calculation were done with commercial software: DISTIL and

HYSYS. The results presented in this work were obtained with UNIQUAC.

3.� �����������������

We follow, basically, the Tao’s procedure (Tao et al., 2003) to generate sequence for ethanol dehydration.

The procedure has two parts. The first part called high purity alternatives is applied in three steps:

1) Characterize the mixture. To begin the problem, one is given a phase equilibrium model, the feed

composition and flow rate, and the column pressure. Then the residue curve map structure is calculated by

computing composition and temperature of all azeotropes, stability of all singular points, liquid-liquid

equilibrium, and distillation regions. Figure 1 displays mixture characteristic for ethanol-benzene-water.

2) Alternatives from decanting and distillation. The sequences were built with simple separation stage using

a set of instructions as guide. Simple separation stage (one feed, two product) is used because we consider a

mixture with a maximum of two liquid phases and simple columns with a single feed and two product streams.

The set of instructions derive from three basic questions: a) Does the stream match a desired exit purity ?, b) Is

the stream heterogeneous ?, and c) Can the stream be split by distillation ?

������ STN diagram corresponding to a sequence of three columns and a decanter.

3) Alternatives from mixing and recycling. The sequence is improved by mixing and recycling working

with stream that are product or nonproduct exit. Here the key question is: Does the mixture of two or more

streams produces a stream with composition in other distillation region or in the heterogeneous region ?

In the second part recycling of streams allows to generate high-purity alternatives. A simple heuristic is used

in this part. The heuristic said that the recycle destination should be chosen so that an exit point for each

component is reachable from the recycle mixing point. In that way all components in a recycled stream could be

separated and they are not accumulated in the recycle loop.

������ STN diagram corresponding to a sequence of two columns and a decanter.

������ Three columns sequence derived from diagram of Fig. 2 (case 1).

The synthesis started with 46000 kg/h of an ethanol-water stream with 0.04 of ethanol molar fraction.

Product specification was an ethanol rich stream of 0.999 of molar fraction allowing very small amount of

benzene and a wastewater stream with an ethanol molar fraction of 0.002 and a very small amount of benzene.

The state-task network (STN), introduced by Sargent (1998), was used for representing the implementation of

the procedure. Following the described synthesis procedure we obtained several sequences. Two of these

sequences are shown in Figure 2 and Figure 3. In the figures streams are vertices and tasks are lines. From the

STN it is easy to obtain process flowsheet as shown in Figures 4 and 5 for the two alternatives generated using

the procedure.

���� � Two columns sequence derived from diagram of Fig. 3 (case 2).

����!� Sequence of two columns (case 3).

4.� ��������"#��$�%�����

Five sequences were optimized. Case 1 (Figure 4) and case 2 (Figure 5) are the sequences generated in item

3. Case 3 (Figure 6), case 4 (Figure 7) and case 5 (Figure 8) were extracted from the literature (patents). All

sequences were feed with the 46000 kg/h ethanol-water stream with 0.04 of ethanol molar fraction. Product

specifications were an ethanol rich stream of 0.999 of molar fraction allowing very small amount of benzene

and a wastewater stream with an ethanol molar fraction of 0.002 and a very small amount of benzene.

����&� Sequence of three columns of Kubierschky (case 4).

����'�Sequence of three columns (case 5).

In the optimization first step we did mass balance for each sequence using logical operation of HYSYS.

Figure 9 shows the computer flowsheet for case 3 (Figure 6). Results from the first step are then used in a second

step where design cases simulated with HYSYS allow to obtain several design alternatives changing column top

vapor composition and column feed composition. For case 3 there are in the literature (Ryan and Doherty, 1989)

recommendations to pick up for distillate composition. Figure 10 shows different design alternatives that are

obtained by changing composition of stream V2 and water molar fraction of stream D3. In particular, case 5

����(�Simulationof two columns sequence of Fig. 6: optimization step 1.

�����)�HYSYS simulation of case 5.

(Figure 8) offers, when the case is compared with other cases, the possibility of modifying composition of the

stream resulting of mixing V2 and D3. For each case at least four alternative designs were analyzed. In a third

step columns in each sequence were designed, in some cases design was done with DISTIL and in others cases

with HYSYS. Figure 10 displays a HYSYS simulation for case 5. In this way some alternatives were eliminated

because it was not possible to find a design that fit specification. This difficulty of finding a feasible design

points out the known fact that heterogeneous azeotropic distillation columns are difficult to operate and control.

�������Different design alternatives for sequence of case 3.

������� Design alternatives for sequence of case 4.

The final results in this step were a total number of column plates and a factor V/F (sequence total vapor flow

divided by feed flow) for all feasible alternatives in each case. These two parameters allow comparison between

sequences. The design alternative of case 3 were done with DISTIL. For some values of V2 molar compositions

it is not possible to find a feasible design. The design for ten D3 molar compositions is presented in Table 1. The

numbers of Table 1 show the tendency that decreasing water concentration in the overall feed to the azeotropic

column (D3) increases the values of the total number of column plates and the V/F factor. Decreasing the

amount of water in D3 places the feed near the distillation border. For ethanol-benzene-water the distillation

border between region I and II is set in an ample range of concentration at a constant water molar fraction as

Figures 1 and 11 show. To reduce design alternatives in other cases we have set the water molar fraction in the

overall feed to the azeotropic column in 0.12. In this way sequences have one degree of freedom less and is

possible to carry out complete sequence mass balances. The water composition restriction was used in case 4.

Figure 12 displays the alternatives analyzed in case 4.

5.� *���������+���������

In �������� ��������� we have produced separation sequences with a branched structure and simple

operations. However, it is possible to propose changes that will generate improvement in these sequences

without violating the set of instructions used. Starting with case 2 (Figure 3) it is possible to:

1) Replace mixture of Sm1 with Sm8 and feeding the preconcentration column with independent feeding

to the column.

2) As the azeotropic column is a stripper using stream Sm8 as reflux in place of mixing with Sm3 allows

Sm3 to feed the azeotropic column in the optimum plate.

,������ Results obtained for the design alternatives of case 3.

Composition of stream D3x ethanol x water

0.100 0.107 0.110 0.120

0.8844 N=90V/F=0.5405

0.8768 N=77V/F=0.4817

0.8730 N=66V/F=0.5875

0.8617 N=59V/F=0.4691

0.8731 N=75V/F=0.6298

0.8599 N=77V/F=0.4330

0.8477 N=66V/F=0.3937

0.8377 N=92V/F=0.5111

0.8204 N=77V/F=0.4388

0.8051 N=67V/F=0.4380

3) Combine the organic and the aqueous streams leaving the decanter to increase azeotropic column reflux.

Doing all the mentioned changes will generate a sequence that is similar to case 3 that, as Table 2 displays, it

is a better sequence than case 2 because it has less total number of plates and a lower value of the V/F factor.

Table 2 summarizes optimization results for the five cases. It is possible to see the great difference in total

number of plates between three columns sequences and two columns sequences. Starting from case 3 the

increase in plate number is accompanied by a decrease in the value of the factor V/F. Comparison shown in

Table 2 help as pattern to see how inconvenient are sequences of cases 1 and 2. Case 2 is an improvement over

case 1, but less suitable in comparison with case 3.

,������ Optimization results.

Case Number of columns Number of steps V/F

1 3 131 0.7923

2 2 91 0.4731

3 2 66 0.3937

4 3 129 0.3476

5 3 141 0.3462

6.� -����������

The procedure to generate sequences was practical in making new separation alternatives for ethanol

dehydration. However, removing some restriction in the procedure, such as restraining to consider only simple

stage of separation, will improve the procedure. The procedure to optimize sequences in spite of having

limitations in its hypothesis and its calculations proves useful to compare sequences at a preliminary stage of

sequence design. Control of the heterogeneous azeotropic column will be critical because of the important

sensitivity and inadequate operational flexibility of sequences.

*�.�������

Bekiaris, N., Meski, G.A., Morari, M. (1996). Multiple Steady States in Heterogeneous Azeotropic Distillation. ��������.����� ������ 207.

Bekiaris, N., Güttinger, T.E., Morari, M. (2000). Multiple Steady States in Distillation: Effect of VL(L)E Inaccuracies.�����������, 955.

Feng, G., Fan, L.T., Friedler, F., Seib, P.A. (2000). Identifying Operating Units for the Design and Synthesis of Azeotropic-

Distillation Systems. ��������������� ������� 175.

Pham, H.N., Doherty, M.F. (1990 a). Design and Synthesis of Heterogeneous Azeotropic Distillation . I. Heterogeneous Phase

Diagrams. ������������������� 1823.

Pham, H.N., Doherty, M.F. (1990 b). Design and Synthesis of Heterogeneous Azeotropic Distillation . II. Residue Curve

Maps, ������������������, 1837.

Pham, H.N., Doherty, M.F. (1990 c). Design and Synthesis of Heterogeneous Azeotropic Distillation . III. Column

Sequences. ������������������, 1845.

Ryan, P.J., Doherty, M.F. (1989). Design / Optimization of Ternary Heterogeneous Azeotropic Distillation Sequences.

�����������, 1592.

Sargent, R.W.H. (1998). Functional Approach to Process Synthesis and Its Application to Distillation Systems. �������

������������� 31.

Tao, L., Malone, M.F., Doherty, M.F. (2003). Synthesis of Azeotropic Distillation Systems with Recycles. Ind. Eng. Chem.

Res., 42, 1783.

Widagdo, S., Seider, W.D. (1996). Azeotropic Distillation. �����������, 96.

��/�� �����$����

The authors are thankful for the financial aid received from CONICET and UNL.