Chapter 8: Introduction to Complex Distillation Methods

We have looked at binary and multicomponent mixtures in both simple and fairly complex columns. However, the chemicals separated have usually had fairly simple equilibrium behavior. In this chapter you will be introduced to a variety of more complex distillation systems used for the separation of less ideal mixtures.

Simple distillation columns are not able to completely separate mixtures when azeotropes occur, and the columns are very expensive when the relative volatility is close to 1. Distillation columns can be coupled with other separation methods to break the azeotrope. This is discussed in the first section. Extractive distillation, azeotropic distillation, and two-pressure distillation are methods for modifying the equilibrium to separate these complex mixtures. These three methods will be described in Sections 2 to 7 of this chapter. In Section 8 we will discuss the use of a distillation column as a chemical reactor, to simultaneously react and separate a mixture.

8.1 Breaking Azeotropes with Other Separators

Azeotropic systems normally limit the separation that can be achieved. For an azeotropic system such as ethanol and water (shown in Figures 2-2 and 4-13), it isn’t possible to get past the azeotropic concentration of 0.8943 mole frac ethanol with ordinary distillation. Some other separation method is required to break the azeotrope. The other method could employ adsorption (Chapter 17), membranes (Chapter 16), extraction (Chapters 13 and 14), and so forth. It could also involve adding a third component to the distillation to give the azeotropic and extractive distillation systems discussed later in this chapter.

Three ways of using an additional separation method to break the azeotrope are shown in Figure 8-1. The simplest, but least likely to be used, is the completely uncoupled system shown in Figure 8-1A. The distillate, which is near the azeotropic concentration, is sent to another separation device, which produces both the desired products. If the other separator can completely separate the products, why use distillation at all? If the separation is not complete, what would be done with the waste stream?

Figure 8-1. Breaking azeotropes; A) separator uncoupled with distillation, B) recycle from separator to distillation, C) recycle and reflux from separator to distillation, D) McCabe-Thiele diagram for part C

A more likely configuration is that of Figure 8-1B. The incompletely separated stream is recycled to the distillation column, which now operates as a two-feed column, so the design procedures used for two-feed columns (Example 4-5) can be used. The arrangement shown in Figure 8-1B is commonly used industrially. The separator may actually be several separators.

An intriguing alternative is the coupled system shown in Figure 8-1C. Now some of the A product is used as reflux to the distillation column. Thus, x₀ > y₁. The McCabe-Thiele diagram for this is illustrated in Figure 8-1D. Note that using a fairly pure reflux stream allows the column to produce a vapor, y₁, which is greater than the azeotropic concentration. This arrangement is explored further in Problem 8-D1.

8.2 Binary Heterogeneous Azeotropic Distillation Processes

The presence of an azeotrope can be used to separate an azeotropic system. This is most convenient if the azeotrope is heterogeneous; that is, the vapor from the azeotrope will condense to form two liquid phases that are immiscible. Azeotropic distillation is often performed by adding a solvent or entrainer that forms an azeotrope with one or both of the components. Before discussing these more complex azeotropic distillation systems in Section 7, let us consider the simpler binary systems that form a heterogeneous azeotrope.

8.2.1 Binary Heterogeneous Azeotropes

Although not common, there are systems such as n-butanol and water, which form a heterogeneous azeotrope (Figure 8-2). When the vapor of the azeotrope with mole frac y_az is condensed, a water-rich liquid phase α in Figure 8-2 and an organic liquid phase β separate from each other. For this type of heterogeneous azeotrope the two-column systems shown in Figure 8-3A can provide a complete separation. Column 1 is a stripping column that receives liquid of composition x_α from the liquid-liquid settler. It operates on the left-hand side of the equilibrium diagram shown in Figure 8-3B. In this region, species B is more volatile and the bottoms from column 1 is almost pure A (x_B,bot1 ~ 0). The overhead vapor from column 1, y₁¹, is condensed and then goes to the liquid-liquid settler where it separates into two liquid phases. Liquid of composition x_α is refluxed to column 1, while liquid of composition x_β is refluxed to column 2.

Figure 8-2. Heterogeneous azeotrope system, n-butanol and water at 1 atmosphere (Perry et al., 1963, p. 13-4).

Figure 8-3. Binary heterogeneous azeotrope; A) two-column distillation system, B) McCabe-Thiele diagram

The second column operates on the right-hand side of Figure 8-3B, where species B is the less volatile component. Thus, the bottoms from this column is almost pure B (x_B,bot2 ~ 1.0). The overhead vapor, which is richer, in species A, is condensed and sent to the liquid-liquid separator.

The liquid-liquid separator takes the condensed liquids, x_α < x < x_β, and separates it into the two liquid phases in equilibrium at mole fracs x_α and x_β. These liquids are used as reflux to columns 1 and 2. The liquid-liquid separator allows one to get past the azeotrope and is therefore a necessary part of the equipment.

The overall external mass balance for the two-column system shown inFigure 8-3A is

(8-1)

while the external mass balance on component B is

(8-2)

Solving these equations simultaneously for the unknown bottoms flow rates, we obtain

(8-3)

(8-4)

Note that this result does not depend on the details of the distillation system.

Analysis of stripping column 1 is also straightforward. The bottom operating equation is

(8-5)

The feed to this column is the saturated liquid reflux of composition x_B,α. This is a vertical feed line. Then the overhead vapor y_B,1¹ is found on the operating line at x_B,α.

The bottom operating equation for column 2 is

(8-6)

The top operating line is a bit different. The easiest mass balance to write uses the mass balance envelope shown in Figure 8-3A. Then the top operating equation is

(8-7)

This is somewhat unusual, because it includes a bottoms concentration leaving the first column. The reflux for this top operating line is liquid of composition x_B,β.

The McCabe-Thiele diagram for this system is shown in Figure 8-3B, where column 2 appears upside down because we have plotted y_B vs. x_B, and B is the less volatile component in column 2. (If this is not clear, return to Problem 4-A4 or turn the diagram upside down and relabel axes as water mole fracs when working with column 2.) The two reflux streams are x_B,α and x_B,β, and reflux is not at the usual value of y_B = x_B = x_B,dist.

The two dashed lines in Figure 8-3B show the route of the overhead vapor streams as they are condensed to saturated liquids (made into x values) and then sent to the liquid-liquid separator. The lever-arm rule [see Eq. (2-26) and Figure 2-10] can be applied to the liquid-liquid separator (see Example 8-1).

Several modifications of the basic arrangement shown in Figure 8-3A can be used. If the feed composition is less than x_B,α, then column 1 should be a complete column and column 2 would be just a stripping column (see Problem 8-C2). The liquids may be subcooled so that the liquid-liquid separator operates below the boiling temperature. This can be advantageous, since the partial miscibility of the system depends on temperature. When the liquids are subcooled, the separator calculation must be done at the temperature of the settler. Then the reflux concentrations can be plotted on the McCabe-Thiele diagram. Design of separators (decanters) is discussed by Woods (1995). More details for heterogeneous azeotropes are explored by Doherty and Malone (2001), Hoffman (1964), and Shinskey (1984). Single-column configurations for binary heterogeneous azeotropes similar to those shown in Figure 8-4 are discussed by Luyben (1973). These configurations are illustrated in Example 8-1.

Figure 8-4. Distillation to dry organic that is partially miscible with water

8.2.2 Drying Organic Compounds That Are Partially Miscible with Water

For partially miscible systems of organics and water, a single phase is formed only when the water concentration is low or very high. For example, a small amount of water can dissolve in gasoline. If more water is present, two phases will form. In the case of gasoline, the water phase is detrimental to the engine, and in cold climates it can freeze in gas lines, immobilizing the car. Since the solubility of water in gasoline decreases as the temperature is reduced, it is important to have dry gasoline.

Fortunately, small amounts of water can easily be removed by distillation or adsorption (Chapter 17). During distillation the water acts as a very volatile component, so a mixture of water and organics is taken as the distillate. After condensation, two liquid phases form, and the organic phase can be refluxed. The system is a type of heterogeneous azeotropic system similar to those discussed in the previous section. Drying differs from the previous systems since a pure water phase is usually not desired, the relative solubilities are often quite low, and the water phase is usually sent to waste treatment. Thus, the system will look like Figure 8-4 with a single column and a phase separator. With very high relative volatilities, one equilibrium stage may be sufficient and a flash system plus a separator can be used.

Simplified equilibrium theories are useful for partially immiscible liquids. There is always a range of concentrations where the species are miscible even though the concentrations may be quite small. For the water phase, it is reasonable as a first approximation to assume that the water follows Raoult’s law and the organic components follow Henry’s law (Robinson and Gilliland, 1950). Thus,

(8-8)

where H_org is the Henry’s law constant for the organic component in the aqueous phase, VP_w is the vapor pressure of water, and x_{w in w} and x_{org in w} are the mole fracs of water and organic in the water phase, respectively. In the organic phase, it is reasonable to use Raoult’s law for the organic compounds and Henry’s law for the water.

(8-9)

where H_w is the Henry’s law constant for water in the organic phase. At equilibrium, the partial pressure of water in the two phases must be equal. Thus, equating p_w in Eqs. (8-8) and (8-9) and solving for H_w we obtain

(8-10)

Similar manipulations for the organic phase give

(8-11)

Using Eqs. (8-10) and (8-11), we can calculate the Henry’s law constants from the known solubilities (which give the mole fracs) and the vapor pressures.

Eqs. (8-8) to (8-11) are valid for both drying organic compounds and steam distillation. The ease of removing small amounts of water from an organic compound that is immiscible with water can be seen by estimating the relative volatility of water in the organic phase.

(8-12)

In the organic phase Eqs. (8-9) and (8-10) can be substituted into Eq. (8-12) to give

(8-13)

This calculation is illustrated in Example 8-1.

If data for the heterogeneous azeotrope (y_w and x_{w in org}) is available, then y_org = 1—y_w, x_{org in org} = 1—x_{w in org}, and α_{w–org in org} is easily estimated from the data [the first equal sign in Eq. (8-12)]. This will be more accurate than assuming Raoult’s law.

Organics can be dried either by continuous distillation or by batch distillation. In both cases the vapor will condense into two phases. The water phase can be withdrawn and the organic phase refluxed to the distillation system. For continuous systems, the McCabe-Thiele design procedure can be used. The McCabe-Thiele graph can be plotted from Eq. (8-13), and the analysis is the same as in the previous section. This is illustrated in Example 8-1.

EXAMPLE 8-1. Drying benzene by distillation

A benzene stream contains 0.01 mole frac water. Flow rate is 1000 kg moles/hr, and feed is a saturated liquid. Column has saturated liquid reflux of the organic phase from the liquid-liquid separator (see Figure 8-4) and uses L/D = 2 (L/D)_min. We want the outlet benzene to have x_{w in benz,bot} = 0.001. Design the column and the liquid-liquid settler.

Solution

A . Define.The column is the same as Figure 8-4. Find the total number of stages and optimum feed stage. For the settler determine the compositions and flow rates of the water and organic phases.

B . Explore. Need equilibrium data. From Robinson and Gilliland (1950), x_{benz in w} = 0.00039, x_{w in benz} = 0.015. This solubility data gives the compositions of the streams leaving the settler. Use Eq. (8-13) for equilibrium. At the boiling point of benzene (80.1°C), VP_benz = 760 mmHg and VP_w = 356.6 mmHg (Perry and Green, 1997). Operation will be at a different temperature, but the ratio of vapor pressures will be approximately constant.

C . Plan. Calculate equilibrium from Eq. (8-13):

This is valid for x_{w in benz} < 0.015. After that, we have a heterogeneous azeotrope. Plot the curve represented by this value of α_w-benz on a McCabe-Thiele diagram. (Two diagrams will be used for accuracy.) Solve with the McCabe-Thiele method as a heterogeneous azeotrope problem. Mass balances can be used to find flow rates leaving the settler.

D . Do it. Plot equilibrium:

where y and x are mole fracs of water in the benzene phase. This is valid for x_w ≤ 0.015. See Figure 8-5A. Since Figure 8-5a is obviously not accurate, we use Figure 8-5B. Calculate vapor mole frac in equilibrium with feed,

Figure 8-5. Solution for Example 8-1; (A) McCabe-Thiele diagram for entire range, (B) McCabe-Thiele diagram for low concentrations

Top Operating Line:

where x_D is water mole frac in distillate = 1—x_{benz in w} = 1—0.00039

y intercept (x = 0) = 0.132

y = x = x_D = 0.99961

Plot top operating line.

Feed Line: Saturated liquid.

Bottom Operating Line:

goes through y = x = x_B = 0.001 and intersection of top operating line and feed line.

Reflux is the benzene phase from the liquid-liquid separator (see Figure 8-4); thus, x_reflux = x_{w in benz} = 0.015. Use this to start stepping off stages. Optimum feed stage is top stage of column. We need 2 stages plus a partial reboiler.

Settler mass balance requires distillate flow rate. From Eq. (3-3),

E. Check. All of the internal consistency checks work. The value of α_w–benz agrees with the calculation of Robinson and Gilliland (1950). The best check on α_w–benz would be comparison with data.

A check on the settler flow rates can be obtained from a water mass balance.

The vapor leaving Stage 1, y_1,w, is a passing stream to the reflux,

y_1,w = (0.868)(0.015) + (0.132)(0.99961) = 0.145

This stream is condensed and separated into the water layer (distillate) and the organic layer (reflux).

Is V₂y_1,w = Dx_D,w + Lx_reflux,w?

(68.55) (0.145) = (9.01) (0.99961) + (59.54) (0.015)?

9.94 = 9.90, which is within the accuracy of the graph.

Generalize. Since the solubility of organics in water is often very low, this type of heterogeneous azeotrope system requires only one distillation column.

Even though water has a higher boiling point than benzene, the relative volatility of water dissolved in benzene is extremely high. This occurs because water dissolved in an organic cannot hydrogen bond as it does in an aqueous phase, and thus, it acts as a very small molecule that is quite volatile. The practical consequence of this is that small amounts of water can easily be removed from organics if the liquids are partially immiscible. There are alternative methods for drying organics such as adsorption that may be cheaper than distillation in many cases.

8.3 Steam Distillation

In steam distillation, water (as steam) is intentionally added to the distilling organic mixture to reduce the required temperature and to keep suspended any solids that may be present. Steam distillation may be operated with one or two liquid phases in the column. In both cases the overhead vapor will condense into two phases. Thus, the system can be considered a type of azeotropic distillation where the added solvent is water and the separation is between volatiles and nonvolatiles. This is a pseudo-binary distillation with water and the volatile organic forming a heterogeneous azeotrope. Steam distillation is commonly used for purification of essential oils in the perfume industry, for distillation of organics obtained from coal, for hydrocarbon distillations, and for removing solvents from solids in waste disposal (Ellerbe, 1997; Ludwig, 1997; Woodland, 1978).

For steam distillation with a liquid water phase present, both the water and organic layers exert their own vapor pressures. At 1 atm pressure the temperature must be less than 100°C even though the organic material by itself might boil at several hundred degrees. Thus, one advantage of steam distillation is lower operating temperatures. With two liquid phases present and in equilibrium, their compositions will be fixed by their mutual solubilities. Since each phase exerts its own vapor pressure, the vapor composition will be constant regardless of the average liquid concentration. A heterogeneous azeotrope is formed. As the amount of water or organic is increased, the phase concentrations do not change; only the amount of each liquid phase will change. Since an azeotrope has been reached, no additional separation is obtained by adding more stages. Thus, only a reboiler is required. This type of steam distillation is often done as a batch operation (see Chapter 9).

Equilibrium calculations are similar to those for drying organics except that now two liquid phases are present. Since each phase exerts its own partial pressure, the total pressure is the sum of the partial pressures. With one volatile organic,

(8-14)

Substituting in Eqs. (8-8) and (8-9), we obtain

(8-15)

The compositions of the liquid phases are set by equilibrium. If total pressure is fixed, then Eq. (8-15) enables us to calculate the temperature. Once the temperature is known the vapor composition is easily calculated as

(8-16)

The number of moles of water carried over in the vapor can be estimated, since the ratio of moles of water to moles organic is equal to the ratio of vapor mole fracs.

(8-17)

Substituting in Eq. (8-16), we obtain

(8-18)

If several organics are present, y_org and p_org are the sums of the respective values for all the organics. The total moles of steam required is n_w plus the amount condensed to heat and vaporize the organic.

EXAMPLE 8-2. Steam distillation

A cutting oil that has approximately the properties of n-decane (C₁₀ H₂₂) is to be recovered from nonvolatile oils and solids in a steady-state single-stage steam distillation. Operation will be with liquid water present. The feed is 50 mole % n-decane. A bottoms that is 15 mole % n-decane in the organic phase is desired. Feed rate is 10 kg moles/hr. Feed enters at the temperature of the boiler. Pressure is atmospheric pressure, which in your plant is approximately 745 mmHg. Find:

a . The temperature of the still

b . The moles of water carried over in the vapor

c . The moles of water in the bottoms

Solution

A.Define. The still is sketched in the figure. Note that there is no reflux.

B.Explore. Equilibrium is given by Eq. (8-15). Assuming that the organic and water phases are completely immiscible, we have in the bottoms x_{C10 in org} = 0.15, x_{w in w} = 1.0 and in the two distillate layers x_C10,org,dist = 1.0 and x_w,water,dist = 1.0.Vapor pressure data as a function of temperature are available in Perry and Green (1997). Then Eq. (8-15) can be solved by trial and error to find T_boiler. Equation (8-18) and a mass balance can be used to determine the moles of water and decane vaporized. The moles of water condensed to vaporize the decane can be determined from an energy balance. Latent heat data are available in Perry and Green (1997).

C. Plan. On a water-free basis the mass balances around the boiler are

(8-19a)

(8-19b)

where B is the bottoms flow rate of the organic phase. Since F = 10, x_bot = 0.15, and z_C10 = 0.5, we can solve for n_C10,vapor and B. Eq. (8-18) gives n_w once T_boiler is known. Since the feed, bottoms, and vapor are all at T_boiler, the energy balance simplifies to

n_C10λ_C10 = (moles of water condensed in still)λ_w

D. Do it. a. Perry and Green (1997) give the following n-decane vapor pressure data (vapor pressure in mm Hg and T in°C):

A very complete table of water vapor pressures is given in that source (see Problem 8-D10). As a first guess, try 95.5°C, where (VP)_w = 645.7 mmHg. Then Eq. (8-15),

(VP)_C10 x_C10 + (VP)_w x_w = p_tot

becomes

(60)(0.15) + (645.7)(1.0) = 654.7 < 745 = p_tot

where we have assumed completely immiscible phases so that x_w = 1.0. This temperature is too low. Approximate solution of Eq. (8-15) eventually gives T_boiler = 99°C.

b. Solving the mass balances, Eqs. (8-19), the kg moles/hr of vapor are

(8-20)

which is 586.2 kg/hr. Eq. (8-18) becomes

or

which is 5347.1 kg/hr.

c. The moles of water to vaporize the decane is

where λ = H_vap—h_liq and the saturated vapor and liquid enthalpies at 99°C are interpolated from Tables 2-249 and 2-352 in Perry and Green (1997) for decane and water, respectively.

E. Check. A check for complete immiscibility is advisable since all the calculations are based on this assumption.

F. Generalize. Obviously, the decane is boiled over at a temperature well below its boiling point, but a large amount of water is required. Most of this water is carried over in the vapor. On a weight basis, the kilograms of total water required per kilogram decane vaporized is 9.26. Less water will be used if the boiler is at a higher temperature and there is no liquid water in the still. Less water is also used for higher values of x_org,bot (see Problem 8-D10).

Additional separation can be obtained by operating without a liquid water phase in the column. Reducing the number of phases increases the degrees of freedom by one. Operation must be at a temperature higher than that predicted by Eq. (8-15), or a liquid water layer will form in the column. Thus, the column must be heated with a conventional reboiler and/or the sensible heat available in superheated steam. The latent heat available in the steam cannot be used, because it would produce a layer of liquid water. Operation without liquid water in the column reduces the energy requirements but makes the system more complex.

8.4 Two-Pressure Distillation Processes

Pressure affects vapor-liquid equilibrium (VLE), and in systems that form azeotropes it will affect the composition of the azeotrope. For example, Table 2-1 shows that the ethanol-water system has an azeotrope at 0.8943 mole frac ethanol at 1 atm pressure. If the pressure is reduced, the azeotropic concentration increases (Seader, 1984). At pressures below 70 mmHg, the azeotrope disappears entirely, and the distillation can be done in a simple column. Unfortunately, use of this disappearance of the azeotrope for the separation of ethanol and water is not economical because the column requires a large number of stages and has a large diameter (Black, 1980). However, the principle of finding a pressure where the azeotrope disappears may be useful in other distillations. The effect of pressure on the azeotropic composition and temperature can be estimated using the VLE correlations in process simulators.

Even though the azeotrope may not disappear, in general, pressure affects the azeotropic composition. If the shift in composition is large enough, a two-column process using two different pressures can be used to completely separate the binary mixture. A schematic of the flowchart for this two-pressure distillation process is shown in Figure 8-6 (Doherty and Malone, 2001; Frank, 1997; Drew, 1997; Shinskey, 1984; Van Winkle, 1967). Column 1 usually operates at atmospheric pressure, while column 2 is usually at a higher pressure but can be at a lower pressure.

Figure 8-6. Two-pressure distillation for azeotropic separation

To understand the operation of this process, consider the separation of methyl ethyl ketone (MEK) and water (Drew, 1997). At 1 atm the azeotrope contains 35% water, while at 100 psia the azeotrope is 50% water. If a feed containing more than 35% water is fed to the first column, the bottoms will be pure water. The distillate from this atmospheric column will be the 35% azeotrope. When this azeotrope is sent to the high-pressure column, an azeotrope containing 50% water comes off as the distillate; this distillate is recycled to column 1. Since the feed to column 2 (the 35% azeotrope) contains less water than this distillate, the bottoms from column 2 is pure MEK. Note that the water is less volatile in column 1 and the MEK is less volatile in column 2. The McCabe-Thiele diagram for one of the columns will have the equilibrium curve and the operating lines below the y = x line.

Mass balances for the system shown in Figure 8-6 are of interest. The external mass balances are identical to Eqs. (8-1) and (8-2). Thus, the bottoms flow rates are given by Eqs. (8-3) and (8-4). Although the processes shown in Figures 8-3A and 8-6 are very different, they look the same to the external mass balances. Differences in the processes become evident when balances are written for individual columns. For instance, for column 2 in Figure 8-6 the mass balances are

(8-21)

and

(8-22)

Solving these equations simultaneously and then inserting the results in Eq. (8-4), we obtain

(8-23)

This is of interest since D₂ is the recycle flow rate. As the two azeotrope concentrations at the two different pressures approach each other, x_dist1—x_dist2 will become small. According to Eq. (8-23), the recycle flow rate D₂ becomes large. This increases both operating and capital costs and makes this process too expensive if the shift in the azeotrope concentration is small.

The two-pressure system is also used for the separation of acetonitrile-water, tetrahydrofuran-water, methanol-MEK, and methanol-acetone (Frank, 1997). In the latter application the second column is at 200 torr. Realize that these applications are rare. For most azeotropic systems the shift in the azeotrope with pressure is small, and use of the system shown in Figure 8-6 will involve a very large recycle stream. This causes the first column to be rather large, and costs become excessive.

Before the relatively recent development of detailed and accurate VLE correlations, most VLE data was only available at one atmosphere; thus, many azeotropic systems have probably not been explored as candidates for two-pressure distillation. Fortunately, it is fairly easy to simulate two-pressure distillation with a process simulator (see Lab 7, part A in the appendix to Chapter 8). Methods for estimating VLE and rapidly screening possible systems are available (Frank, 1997). If two-pressure distillation were routinely considered as an option for breaking azeotropes, we would undoubtedly discover additional systems where this method is economical.

8.5 Complex Ternary Distillation Systems

In Chapters 5, 6 and 7 we studied multicomponent distillation for systems with relatively ideal VLE that do not exhibit azeotropic behavior. In Chapter 4 when we studied both relatively ideal and azeotropic binary systems we found that there were significant differences between these systems. If no azeotrope forms, one can obtain essentially pure distillate and pure bottom products. If there is an azeotrope we found that one can at best obtain one pure product and the azeotrope. In Section 8.2 we found that if the binary azeotrope was heterogeneous one could usually use a liquid-liquid separator to get past the azeotrope and obtain two pure products with two columns. Ternary systems with non-ideal VLE can have one or more azeotropes that may be homogeneous or heterogeneous. Since the behavior of ideal ternary distillation is more complex than that of ideal binary distillation, we expect that the behavior of nonideal ternary distillation is probably more complex than nonideal binary distillation.

Although McCabe-Thiele diagrams can be used for ternary systems, they have not been nearly as successful as the binary applications. Hengstebeck (1961) developed a pseudo-binary approach that is useful for systems with close to ideal VLE. It has also been applied to extractive distillation by assuming the solvent concentration is constant. Chambers (1951) developed a method that could be applied with fewer assumptions to systems with azeotropes and illustrated it with the ternary system methanol-ethanol-water. His approach consisted of drawing two McCabe-Thiele diagrams (e.g., one for methanol and one for ethanol). Equilibrium consists of several curves with methanol mole frac as a parameter on the ethanol diagram and ethanol mole frac as a parameter on the methanol diagram. Each equilibrium step required simultaneous solution of the two diagrams (see Wankat, 1981). The operating lines plot on these diagrams in the normal fashion. Although visually instructive, Chambers’ method is awkward and has not been widely used. The conclusion is we need new visualization tools to study ternary distillation.

8.5.1 Distillation Curves

You may have noticed in Chapter 2 that enthalpy-composition and temperature-composition diagrams contain more information than the McCabe-Thiele y-x diagram. We started using the diagram with less information because it was easier to show the patterns and visualize the separation. For ternary distillation we will repeat this pattern and go to a ternary composition diagram that shows the paths taken by liquid mole fracs throughout a distillation operation. We gain in visualization power since a variety of possible paths are easy to illustrate, but we lose power since the stages are no longer shown. For complex systems the gains are much more important than the losses.

A distillation curve is a plot of the mole fracs on every tray for distillation at total reflux. The two different formats used for these diagrams are shown in Figures 8-7 and 8-8. To generate these plots consider a distillation column numbered from the top down (Figure 6-2). If we start at the reboiler, we first do a bubble point calculation,

(8-24)

Calculation of the bubble point at every stage can be laborious; fortunately, the calculation is easily done with a process simulator. Next, at total reflux the operating equation is,

8-25a

Alternation of these two equations results in values for x_A, x_B, and x_C on every stage. These values are then plotted in Figures 8-7 and 8-8. The starting mole fracs are chosen so that the distillation curves fill the entire space of the diagrams. If the relative volatility is constant, then the vapor mole fracs can be easily calculated from Eq. (6-15b). Substituting Eq. (6-15b) into Eq. (8-25a) and solving for x_i,j-1, we obtain the recursion relationship for the distillation curve for constant relative volatility systems.

8-25b

Figure 8-7. Distillation curves for constant relative volatility system; A = benzene, B = toluene, C = cumene; α_AB = 2.4, α_BB = 1.0, α_BC = 0.21; stages are shown as ×.

Figure 8-8. Distillation curves for acetone, chloroform, benzene mixtures at 1.0 atm (Biegler et al., 1997), reprinted with permission of Prentice-Hall PTR, copyright 1997, Prentice-Hall.

Figure 8-7 shows the characteristic pattern of distillation curves for ideal or close to ideal VLE with no azeotropes. All of the systems considered in Chapters 5, 6, and 7 follow this pattern. The y-axis (x_B = 0) represents the binary A-C separation. This starts at the reboiler (x_A = 0.01 is an arbitrary value) and requires only the reboiler plus 4 stages to reach a distillate value of x_A = 0.994. The x axis (x_A = 0) represents the binary B-C separation, which was started at the arbitrary value x_B = 0.01 in the reboiler. The maximum in B concentration should be familiar from the profiles shown in Chapter 5. Distillation curves at finite reflux ratios are similar but not identical to those at total reflux. Note that the entire space of the diagram can be reached by starting with concentrations near 100% C (the heavy boiler). Distillation curves are usually plotted as smooth curves—they were plotted as discrete points in this diagram to emphasize the location of the stages. The arrows are traditionally shown in the direction of increasing temperature. Your understanding of the procedure for plotting these curves will be aided significantly be doing problem 8.D12.

Figure 8-8 (Biegler et al., 1997) shows the total reflux distillation curves for acetone, chloroform, benzene distillation on an equilateral triangle diagram. This system has a maximum boiling azeotrope between acetone (x_acetone ~ 0.34) and chloroform. To reach compositions on the right side of the diagram we need to start with locations that are to the right of distillation boundary curve c. For compositions on the left side of the diagram we have to start with locations that are to the left of the distillation boundary. For a given feed concentration, only part of the space in Figure 8-8 can be reached for distillate and bottoms products. If we want to produce pure acetone in a single column, the feed needs to be to the left of the distillation boundary. To produce pure chloroform, the feed needs to be to the right of the distillation boundary. Although the distillation boundary will move slightly at finite reflux ratios, this basic principle still holds. Figure 8-8 is unusual because it shows a relatively rare maximum boiling azeotrope. All of the other systems we will consider are the much more common minimum boiling azeotropes.

We can also do mass balances on the triangular diagrams. First, consider the separation of binaries, which occur along the y-axis, the x-axis, and the hypotenuse of the right triangle. In Figure 8-9 the binary separation of the heavy (highest boiling) component H from the intermediate component I occurs along the y-axis (line B₁F₁D₁).

Figure 8-9. Mass balance on triangular diagram

In Chapter 3 we developed Eq. (3-3) for binary separations. This equation applies to each component in the binary separations. For example, for separation of intermediate and heavy components,

(8-26)

Equation (8-26) remains valid if a third (or fourth or more) components are present in the distillation, and we can write a similar equation for every component. Thus, Eq. (8-26) applies for the general ternary separation shown as line BFD. This equation also proves that the points representing bottoms, distillate and feed all lie on a straight line and that the lever-arm rule applies. [This is very similar to the graphical solution developed for the enthalpy-composition diagram of Figure 2-9 and the lever-arm rule derived in Eq. (2-26). The rules for mass balances on triangular diagrams will be developed in detail in Section 14.2.] As a first approximation, the points representing the bottoms and distillate products from a distillation column with a single feed will lie on the distillation and residue curves. Thus, these curves will show us if the separation indicated by the line BFD is feasible. If there is a distillation boundary as in Figure 8-8, not all separations will be feasible.

8.5.2 Residue Curves

We could do all of our calculations with distillation curves at total and finite reflux ratios; however, these curves depend to some extent on the distillation system. It is convenient to use a thermodynamically based curve that does not depend on the number of stages. A residue curve is generated by putting a mixture in a still pot and boiling it without reflux until the pot runs dry and only the residue remains. This is a simple batch distillation that will be discussed in more detail in Chapter 9. The plot of the changing mole fracs on a triangular diagram is the residue curve. It will be similar but not identical to the distillation curves shown in Figures 8-7 and 8-8. The differences between distillation curves and residue curves are explored by Widagdo and Seider (1996).

A simple equilibrium still is shown in Figure 8-10. As the distillation continues the molar holdup of liquid H decreases. The unsteady state overall mass balance is,

8-27a

where V is the molar rate (not necessarily constant) at which vapor is removed. The component mass balances will have a similar form.

Figure 8-10. Simple equilibrium still

8-27b

Expanding the derivative, substituting in Eq. (8-27a) and rearranging, we obtain

(8-28)

This derivation closely follows the derivations in Biegler et al. (1997) and Doherty and Malone (2001). An alternate derivation is given in sections 9.1 and 9.2.

Integration of Eq. (8-28) gives us the values of x_i vs. time and allows us to plot the residue curve. This integration can be done with any suitable numerical integration technique; however, the vapor mole frac y in equilibrium with x must be determined at each time step. Although Doherty and Malone (2001) recommend the use of either Gear’s method or a fourth order Runge-Kutta integration, they note that Eq. (8-28) is well behaved and can be integrated with Euler’s method. This result is particularly simple,

(8-29)

where k refers to the step number and h is the step size. A step size of h = 0.01 or smaller is recommended (Doherty and Malone, 2001). If relative volatilities are constant, we can determine y from Eq. (6-15b) and the recursion relationship simplifies to

(8-30)

Despite the fact that process simulators will do these calculations for us, doing the integration for a simple case will greatly increase your understanding. Thus, problem 8.D13 is highly recommended.

Siirola and Barnicki (1997) show simplified residue curve plots for all 125 possible systems. The most common residue curve is the plot for ideal distillation, which is similar to Figure 8-7. Next most common will be systems with a single minimum boiling azeotrope occurring between one of the sets of binary pairs. The three possibilities are shown in Figure 8-11 (Doherty and Malone, 2001). Figure 8-11c is of interest since it is one of the two residue curves that occur in extractive distillation (the other is Figure 8-7 with small relative volatilities). As mentioned earlier, a residue curve plot for a maximum boiling azeotrope as shown in Figure 8-8 will be rare. One can also have multiple binary and ternary azeotropes (see Siirola and Barnecki, 1997). Heterogeneous ternary azeotropes can also occur and are important in azeotropic distillation (Section 8.7). Figure 8-12 (Doherty and Malone, 2001) is an example of the residue curves that occur in azeotropic distillation with added solvent. The systems shown in Figures 8-11c and 8-12 are often formed on purpose by adding a solvent to a binary azeotropic system.

Figure 8-11. Schematics of residue curve maps when there is one binary minimum-boiling azeotrope (Doherty and Malone, 2001); reprinted with permission of McGraw-Hill, copyright 2001, McGraw-Hill.

Figure 8-12. Calculated residue curve map for ethanol-water-benzene (Doherty and Malone, 2001). The black squares are binary azeotropes, the black triangle is a heterogeneous ternary azeotrope, and the dot-dash line represents the solubility envelope for the two liquid layers. Reprinted with permission of McGraw-Hill, copyright 2001, McGraw-Hill.

This completes the introduction to residue curves. Residue curves will be used in the explanation of extractive distillation (Section 8.6) and azeotropic distillation (Section 8.7). In Section 11.6 we will use residue curves to help synthesize distillation sequences for complex systems. Doherty and Malone (2001) develop the properties and applications of residue curves in much more detail than can be done in this introduction.

8.6 Extractive Distillation

Extractive distillation is used for the separation of azeotropes and close-boiling mixtures. In extractive distillation, a solvent is added to the distillation column. This solvent is selected so that one of the components, B, is selectively attracted to it. Since the solvent is usually chosen to have a significantly higher boiling point than the components being separated, the attracted component, B, has its volatility reduced. Thus, the other component, A, becomes relatively more volatile and is easy to remove in the distillate. A separate column is required to separate the solvent and component B. The residue curves, after the solvent is added, are shown in Figure 8-7 (separation of close boiling components) and 8-11C (separation of azeotropes).

A typical flowsheet for separation of a binary mixture is shown in Figure 8-13. If an azeotrope is being separated, the feed should be close to the azeotrope concentration; thus, a binary distillation column (not shown) usually precedes the extractive distillation system. In column 1 the solvent is added several stages above the feed stage and a few stages below the top of the column. In the top section, the relatively nonvolatile solvent is removed and pure A is produced as the distillate product. In the middle section, large quantities of solvent are present and components A and B are separated from each other. It is common to use 1, 5, 10, 20, or even 30 times as much solvent as feed; thus, the solvent concentration in the middle section is often quite high. Note that the A-B separation must be complete in the middle section, because any B that gets into the top section will not be separated (there is very little solvent present) and will exit in the distillate. The bottom section strips the A from the mixture so that only solvent and B exit from the bottom of the column.

Figure 8-13. Extractive distillation flowsheet

The residue curve diagram for the extractive distillation of ethanol and water using ethylene glycol as the solvent is shown in Figure 8-14. The curves start at the binary azeotrope (89% ethanol and 11% water), which makes this residue curve plot similar to Figure 8-11c. Since all of the residue curves have the same general shape, there is no distillation boundary in this system. (I suggest that you compare the shape of the residue curves in Figure 8-14 to those in Figure 8-8, which has a distillation boundary.) The feed to column 1 is 100 kmoles/hr of a mixture that is 72 mole % ethanol and 27 mole % water. Essentially pure solvent is added at a rate of 52 kmoles/hr. (Lab 8 in the Appendix to Chapter 8.) The solvent and feed can be combined as a mixed feed M (found from the lever-arm rule), which is then separated into streams D and B as shown in the figure. Although the residue curves were not generated for a two feed column, Figure 8-14, allows us to determine that the process is feasible. Detailed simulations are needed to determine if the solvent flow rate and reflux ratio are large enough. Increasing the solvent flow rate will move point M towards the solvent vertex. Binary stream B is easily separated in Column 2. Although this system can be used to break the ethanol-water azeotrope, azeotropic distillation (see Section 8.7) is more economical. Doherty and Malone (2001) provide detailed information on the use of residue curve maps to design extractive distillation systems.

Figure 8-14. Residue curves for water-ethanol-ethylene glycol for extractive distillation to break ethanol-water azeotrope. Curves generated by Aspen Plus 2004 using NRTL for VLE. Mass balance lines FMS and BMD were added.

The mixture of solvent and B are sent to column 2 where they are separated. If the solvent is selected correctly, the second column can be quite short, since component B is significantly more volatile than the solvent. The recovered solvent can be cooled and stored for reuse in the extractive distillation column. Note that the solvent must be cooled before entering column 1, since its boiling point is significantly higher than the operating temperature of column 1.

Column 2 is a simple distillation that can be designed by the methods discussed in Chapter 4. Column 1 is considerably more complex, but the bubble-point matrix method discussed in Chapter 6 can often be adapted. Since the system is nonideal and K values depend on the solvent concentration, a concentration loop is required in the flowchart shown in Figure 6-1. Fortunately, a good first guess of solvent concentrations can be made. Solvent concentration will be almost constant in the middle section and also in the bottom section except for the reboiler. In the top section of the column, the solvent concentration will very rapidly decrease to zero. These solvent concentrations will be relatively unaffected by the temperatures and flow rates. The K values can be calculated from Eq. (2-35) with the activity coefficients determined from the appropriate VLE correlation. Process simulators are the easiest way to do these calculations (see appendix to Chapter 8).

Concentration profiles for the extractive distillation of n-heptane and toluene using phenol as the solvent are shown in Figure 8-15 (Seader, 1984). The profiles were rigorously calculated using a simultaneous correction method, and activity coefficients were calculated with the Wilson equation. The feed was a mixture of 200 lb moles/hr of n-heptane and 200 lb moles/hr of toluene input as a liquid at 200°F on stage 13. The recycled solvent is input on stage 6 at a total rate of 1200 lb moles/hr. There are a total of 21 equilibrium contacts including the partial reboiler. The high-boiling phenol is attractive to the toluene, since both are aromatics. The heptane is then made more volatile and exits in the distillate (component A in Figure 8-13). Note from Figure 8-15 that the phenol concentration very rapidly decreases above stage 6 and the n-heptane concentration increases. From stages 6 to 12, phenol concentration is approximately constant and the toluene is separated from the heptane. From stages 13 to 20, phenol concentration is again constant but at a lower concentration. This change in the solvent (phenol) concentration occurs because the feed is input as a liquid. A constant solvent concentration can be obtained by vaporizing the feed or by adding some recycled solvent to it. Heptane is stripped from the mixture in stages 13 to 20. In the reboiler, the solvent is nonvolatile compared to toluene. Thus the boilup is much more concentrated in toluene than in phenol. The result is the large increase in phenol concentration seen for stage 21 in Figure 8-15. Concentration profiles for other extractive distillation systems are shown by Robinson and Gilliland (1950), Siirola and Barnicki (1997), and Doherty and Malone (2001).

Figure 8-15. Calculated composition profiles for extractive distillation of toluene and n-heptane; from Seader (1984), reprinted with permission from Perry’s Chemical Engineer’s Handbook, 6th ed, copyright 1984, McGraw-Hill.

The almost constant phenol (solvent) concentrations above and below the feed stage in Figure 8-15 allow for the development of approximate calculation methods (Knickle, 1981). A pseudo-binary McCabe-Thiele diagram can be used for the separation of heptane and toluene. The “equilibrium” curve represents the y-x data for heptane and toluene with constant phenol concentration. The equilibrium curve will have a discontinuity at the feed stage. For the stages above the solvent feed stage, a standard heptane-phenol binary McCabe-Thiele diagram can be used. Knickle (1981) also discusses modifying the Fenske-Underwood-Gilliland approach by modifying the relative volatility to represent heptane-toluene equilibrium with the phenol concentration of the feed stage.

The reason for the large increase in solvent concentration in the reboiler is easily seen if we look at an extreme case where none of the solvent vaporizes in the reboiler. Then the boilup is essentially pure component B. The liquid flow rate in the column can be split up as

(8-31)

where S is the constant solvent flow rate; B_liq is the flow rate of component B, which stays in the liquid in the reboiler; and is the flow rate of the vapor, which is mainly B. The bottoms flow rate consists of the streams that remain liquid,

(8-32)

The mole frac of B in the liquid in the column can be estimated as

(8-33)

and the mole frac B in the bottoms is

(8-34)

For the usual flow rates this gives x_B,col > x_B,bot. Even when the solvent is fairly volatile, as in Figure 8-15, the toluene (component B) concentration drops in the reboiler.

Selection of the solvent is extremely important. The process is similar to that of selecting a solvent for liquid-liquid extraction, which is discussed in Chapter 13. By definition, the solvent should not form an azeotrope with any of the components (Figure 8-11C). If the solvent does form an azeotrope, the process becomes azeotropic distillation, which is discussed in the next section. Usually, a solvent is selected that is more similar to the heavy key. Then the volatility of the heavy key will be reduced. Exceptions to this rule exist; for example, in the n-butane—1-butene system, furfural decreases the volatility of the 1-butene, which is more volatile (Shinskey, 1984). Lists of extractants (Van Winkle, 1967) for extractive distillation are helpful in finding a general structure that will effectively increase the volatility of the keys. Salts can be used as the “solvent,” particularly when there are small amounts of water to remove (Furter, 1993). Residue curve analysis is also very helpful to ensure that the solvent does not form additional azeotropes (Biegler et al., 1997; Doherty and Malone, 2001).

Solvent selection can be aided by considering the polarities of the compounds to be separated. A short list of classes of compounds arranged in order of increasing polarity is given in Table 8-1. If two compounds of different polarity are to be separated, a solvent can be selected to attract either the least polar or the most polar of the two. For example, suppose we wish to separate acetone (a ketone boiling at 56.5°C) from methanol (an alcohol boiling at 64.7°C). This system forms an azeotrope.We could add a hydrocarbon to attract the acetone, but if enough hydrocarbon were added, the methanol would become more volatile. A simpler alternative is to add water, which attracts the methanol and makes acetone more volatile. The methanol and water are then separated in column 2. In this example, we could also add a higher molecular weight alcohol such as butanol to attract the methanol.

Table 8.1. Increasing polarities of classes of compounds

When two hydrocarbons are to be separated, the larger the difference in the number of double bonds the better a polar solvent will work to change the volatility. For example, furfural will decrease the volatility of butenes compared to butanes. Furfural (a cyclic alcohol) is used instead of water because the hydrocarbons are miscible with furfural. A more detailed analysis of solvent selection shows that hydrogen bonding is more important than polarity (Berg, 1969; Smith, 1963).

Once a general structure has been found, homologs of increasing molecular weight can be checked to find which has a high enough boiling point to be easily recovered in column 2. However, too high a boiling point is undesirable, because the solvent recovery column would have to operate at too high a temperature. The solvent should be completely miscible with both components over the entire composition range of the distillation.

It is desirable to use a solvent that is nontoxic, nonflammable, noncorrosive, and nonreactive. In addition, it should be readily available and inexpensive since solvent makeup and inventory costs can be relatively high. Environmental effects and life-cycle costs of various solvents need to be included in the decision (Allen and Shonnard, 2002). As usual, the designer must make tradeoffs in selecting a solvent. One common compromise is to use a solvent that is used elsewhere in the plant or is a by-product of a reaction even if it may not be the optimum solvent otherwise.

For isomer separations, extractive distillation usually fails, since the solvent has the same effect on both isomers. For example, Berg (1969) reported that the best entrainer for separating m- and p-xylene increased the relative volatility from 1.02 to 1.029. An alternative to normal extractive distillation is to use a solvent that preferentially and reversibly reacts with one of the isomers (Doherty and Malone, 2001). The process scheme will be similar to Figure 8-14, with the light isomer being product A and the heavy isomer product B. The forward reaction occurs in the first column, and the reaction product is fed to the second column. The reverse reaction occurs in column 2, and the reactive solvent is recycled to column 1. This procedure is quite similar to the combined reaction-distillation discussed in Section 8.8.

8.7 Azeotropic Distillation with Added Solvent

When a homogeneous azeotrope is formed or the mixture is very close boiling, the procedures shown in Section 8.2 cannot be used. However, the engineer can add a solvent (or entrainer) that forms a binary or ternary azeotrope and use this to separate the mixture. The trick is to pick a solvent that forms an azeotrope that is either heterogeneous (then the procedures of Section 8.2 are useful) or easy to separate by other means such as extraction with a water wash. Since there are now three components, it is possible to have one or more binary azeotropes or a ternary azeotrope. The flowsheet depends upon the equilibrium behavior of the system, which can be investigated with distillation curves and residue curves (Section 8.5). A few typical examples will be illustrated here.

Figure 8-16 shows a simplified flowsheet (extensive heat exchange is not shown) for the separation of butadiene from butylenes using liquid ammonia as the entrainer (Poffenberger et al., 1946). Note the use of the intermediate reboiler in the azeotropic distillation column to minimize polymerization. At 40°C the azeotrope is homogeneous. The ammonia can be recovered by cooling, since at temperatures below 20°C two liquid phases are formed. The colder the operation of the settler the purer the two liquid phases. At the −40°C used in commercial plants during World War II, the ammonia phase contained about 7 wt % butylene. This ammonia is recycled to the azeotropic column either as reflux or on stage 30. The top phase is fed to the stripping column and contains about 5 wt % ammonia. The azeotrope produced in the stripping column is recycled to the separator. This example illustrates the following general points: 1) The azeotrope formed is often cooled to obtain two phases and/or to optimize the operation for the liquid-liquid settler; 2) Streams obtained from a settler are seldom pure and have to be further purified. This is illustrated by the stripping column in Figure 8-16. 3) Product (butylene) can often be recovered from solvent (NH₃) in a stripping column instead of a complete distillation column because the azeotrope is recycled.

Figure 8-16. Separation of butadiene from butylenes using ammonia as an entrainer (Poffenberger et al., 1946)

Another system with a single binary azeotrope is shown in Figure 8-17 (Smith, 1963). In the azeotropic column, component A and the entrainer form a minimum boiling azeotrope, which is recovered as the distillate. The other component, B, is recovered as a pure bottoms product. In this case the azeotrope formed is homogeneous, and a water wash (extraction using water) is used to recover the solvent from the desired component with which it forms an azeotrope. Pure A is the product from the water wash column. A simple distillation column is required to recover the solvent from the water. Chemical systems using flow diagrams similar to this include the separation of cyclohexane (A) and benzene (B), using acetone as the solvent, and the removal of impurities from benzene with methanol as the solvent.

Figure 8-17. Azeotropic distillation with one minimum boiling binary azeotrope; use of water wash for solvent recovery from Smith (1963)

A third example that is quite common is the separation of the ethanol water azeotrope using a hydrocarbon as the entrainer. Benzene used to be the most common entrainer (the residue curve is shown in Figure 8-12), but because of its toxicity it has been replaced by diethyl ether, n-pentane, or n-hexane. A heterogeneous ternary azeotrope is removed as the distillate product from the azeotropic distillation column. A typical flowsheet for this system is shown in Figure 8-18 (Black, 1980; Robinson and Gilliland, 1950; Seader, 1984; Shinskey, 1984; Smith, 1963; Doherty and Malone, 2001; Widagdo and Seider, 1996). The feed to the azeotropic distillation column is the distillate product from a binary ethanol-water column and is close to the azeotropic composition. The composition of the ternary azeotrope will vary slightly depending upon the entrainer chosen. For example, when n-hexane is the entrainer the azeotrope contains 85 wt % hexane, 12 wt % ethanol, and 3 wt % water (Shinskey, 1984). The water/ethanol ratio in the ternary azeotrope must be greater than the water/ethanol ratio in the feed so that all the water can be removed with the azeotrope and excess ethanol can be removed as a pure bottoms product. The upper layer in the separator is 96.6 wt % hexane, 2.9 wt % ethanol and 0.5 wt % water, while the bottom layer is 6.2 wt % hexane, 73.7 wt % ethanol and 20.1 wt % water. The upper layer from the separator is refluxed to the azeotropic distillation column, while the bottom layer is sent to a stripping column to remove water.

Figure 8-18. Ternary azeotropic distillation for separation of ethanol-water with hydrocarbon entrainer

Calculations for any of the azeotropic distillation systems are considerably more complex than for simple distillation or even for extractive distillation. The complexity arises from the obviously very nonideal equilibrium behavior and from the possible formation of three phases (two liquids and a vapor) inside the column. The residue curve shown in Figure 8-12 clearly demonstrates the complexity of these systems. Calculation procedures for azeotropic distillation are reviewed by Prokopakis and Seider (1983) and Widago and Seider (1996).

Results of simulations have been presented by Black (1980), Hoffman (1964), Prokopakis and Seider (1983), Robinson and Gilliland (1950), Seader (1984), Smith (1963), and Widago and Seider (1996). Seader’s (1984) results for the dehydration of ethanol using n-pentane as the solvent are plotted in Figure 8-19. The system used is similar to the flowsheet shown in Figure 8-18. The feed to the column contained 0.8094 mole frac ethanol. The column operated at a pressure of 331.5 kPa to allow condensation of the distillate with cooling water and had 18 stages plus a partial reboiler and a total condenser. The third stage below the condenser was the feed stage. Note that the condenser profiles are different from those shown in Chapter 5. The pentane appears superficially to be a light key except that none of it appears in the bottoms. Instead, a small amount of the water exits in the bottoms with the ethanol.

Figure 8-19. Composition profiles for azeotropic distillation column separating water and ethanol with n-pentane entrainer (Seader, 1984).

Selecting a solvent for azeotropic distillation is often more difficult than for extractive distillation. There are usually fewer solvents that will form azeotropes that boil at a low enough temperature to be easy to remove in the distillate or boil at a high enough temperature to be easy to remove in the bottoms. Distillation curve and residue curve analyses are useful for screening prospective solvents and for developing new processes (Biegler et al., 1997; Doherty and Malone, 2001; Widago and Seider, 1996). In addition, the binary or ternary azeotrope formed must be easy to separate. In practice, this requirement is met by heterogeneous azeotropes and by azeotropes that are easy to separate with a water wash. The chosen entrainer must also satisfy the usual requirements of being nontoxic, noncorrosive, chemically stable, readily available, inexpensive, and green. Because of the difficulty in finding suitable solvents, azeotropic distillation systems with unique solvents are patentable.

8.8 Distillation with Chemical Reaction

Distillation columns are occasionally used as chemical reactors. The advantage of this approach is that distillation and reaction can take place simultaneously in the same vessel, and the products can be removed to drive the reversible reaction to completion. The most common industrial application is for the formation of esters from a carboxylic acid and an alcohol. For example, the manufacture of methyl acetate by reactive distillation was a major success that conventional processes could not compete with (Biegler et al., 1997). Reactive distillation was first patented by Backhaus in 1921 and has been the subject of several patents since then (see Siirola and Barnicki, 1997; and Doherty and Malone, 2001 for references). Reaction in a distillation column may also be undesirable when one of the desired products decomposes.

Distillation with reaction is useful for reversible reactions. Examples would be reactions such as

The purposes of the distillation are to separate the product(s) from the reactant(s) to drive the reactions to the right, and to recover purified product(s).

Depending on the equilibrium properties of the system, different distillation configurations can be used as shown in Figure 8-20. Figure 8-20A shows the case where the reactant is less volatile than the product (Belck, 1955). If several products are formed, no attempt is made to separate them in this system. The bleed is used to prevent the buildup of nonvolatile impurities or products of secondary reactions. If the feed is more volatile than the desired product, the arrangement shown in Figure 8-20B can be used (Belck, 1955). This column is essentially at total reflux except for a small bleed, which may be needed to remove volatiles or gases.

Figure 8-20. Schemes for distillation plus reaction: A) Volatile product, reaction is A = C; B) nonvolatile product, reaction is A = D; C) Two products, reactions are A = C + D or A + B = C + D; (D, E) Reaction A + B = C + D with B and D nonvolatile.

Figures 8-20C, D, and E all show systems where two products are formed and the products are separated from each other and from the reactants in the distillation column. In Figure 8-20C the reactant(s) are of intermediate volatility between the two products. Then the reactants will stay in the middle of the column until they are consumed, while the products are continuously removed, driving the reaction to the right. If the reactants are not of intermediate volatility, some of the reactants will appear in each product stream (Suzuki et al., 1971). The alternative schemes shown in Figures 8-20D and E (Suzuki et al., 1971; Siirola and Barnicki, 1997) will often be advantageous for the reaction

A + B = C + D

In these two figures, species A and C are relatively volatile while species B and D are relatively nonvolatile. Since reactants are fed in at opposite ends of the column, there is a much larger region where both reactants are present. Thus, the residence time for the reaction will be larger in Figures 8-20D and E than in Figure 8-20C, and higher yields can be expected. The systems shown in Figures 8-20C and D have been used for esterification reactions such as

Acetic acid + ethanol » ethyl acetate + water

(Suzuki et al., 1971) and

Acetic acid + methanol » methyl acetate + water

Siirola and Barnicki (1997) show a four-component residue curve map for methyl acetate production. They used a modification of Figure 8-20E where a non-volatile liquid catalyst is fed between the acetic acid (B) and methanol (A) feeds. They show profiles for the four components and the catalyst.

When a reaction occurs in the column, the mass and energy balance equations must be modified to include the reaction terms. The general mass balance equation for stage j (Eq. 6-1) can be modified to

(8-35)

where the reaction term r_j is positive if the component is a product of the reaction. To use Eq. (8-35), the appropriate rate equation for the reaction must be used for r_j. In general, the reaction rate will depend on both the temperature and the liquid compositions.

When the mass balances are in matrix form, the reaction term can conveniently be included with the feed in the D term in Eqs. (6-6) and (6-13). This retains the tridiagonal form of the mass balance, but the D term depends upon liquid concentration and stage temperature. The convergence procedures to solve the resulting set of equations must be modified, because the procedures outlined in Chapter 6 may not be able to converge. The equations have become highly nonlinear because of the reaction rate term. Modern process simulators are usually able to converge for reactive distillation problems, although it may be necessary to change the convergence properties.

A sample of composition and temperature profiles for the esterification of acetic acid and ethanol is shown in Figure 8-21 (Suzuki et al., 1971) for the distillation system of Figure 8-20C. The distillation column is numbered with 13 stages including the total condenser (No. 1) and the partial reboiler (No. 13). Reaction can occur on every stage of the column and in both the condenser and the reboiler. Feed is introduced to stage 6 as a saturated liquid. The feed is mainly acetic acid and ethanol with a small amount of water. A reflux ratio of 10 is used. The top product contains most of the ethyl acetate produced in the reaction plus ethanol and a small amount of water. All of the nonreacted acetic acid appears in the bottoms along with most of the water and a significant fraction of the ethanol. Reaction is obviously not complete.

Figure 8-21. Composition and temperature profiles for the reaction acetic acid + ethanol = ethyl acetate + water from Suzuki et al. (1971), copyright 1971. Reprinted with permission from Journal of Chemical Engineering of Japan.

A somewhat different type of distillation with reaction is “catalytic distillation” (Parkinson, 2005). In this process bales of catalyst are stacked in the column. The bales serve both as the catalyst and as the column packing (see Chapter 10). This process was used commercially for production of methyl tert-butyl ether (MTBE) from the liquid-phase reaction of isobutylene and methanol. The heat generated by the exothermic reaction is used to supply much of the heat required for the distillation. Since MTBE use as a gasoline additive has been outlawed because of pollution problems from leaky storage tanks, these units are shut down. Other applications of catalytic distillation include desulfurization of gasoline, separation of 2-butene from a mixed C₄ stream, and esterification of fatty acids.

Although many reaction systems do not have the right reaction equilibrium or VLE characteristics for distillation with reaction, for those that do this technique is a very valuable industrial tool.

8.9 Summary—Objectives

In this chapter we have looked at azeotropic and extractive distillation systems plus distillation with simultaneous chemical reaction. At the end of this chapter you should be able to satisfy the following objectives:

1. Analyze binary distillation systems using other separation schemes to break the azeotrope

2. Solve binary heterogeneous azeotrope problems, including the drying of organic solvents, using McCabe-Thiele diagrams

3. Explain and analyze steam distillation

4. Use McCabe-Thiele diagrams or process simulators to solve problems where two pressures are used to separate azeotropes

5. Determine the possible products for a ternary distillation using residue curves

6. Explain the purpose of extractive distillation, select a suitable solvent, explain the expected concentration profiles, and do the calculations with a process simulator

7. Use a residue curve diagram to determine the expected products for an azeotropic distillation with an added solvent

8. Explain qualitatively the purpose of doing a reaction in a distillation column, and discuss the advantages and disadvantages of the different column configurations

References

Allen, D. T. and D. R. Shonnard, Green Engineering: Environmentally Conscious Design of Chemical Processes, Prentice-Hall PTR, Upper Saddle River, New Jersey, 2002.

Belck, L. H., “Continuous Reactions in Distillation Equipment,” AIChE J., 1, 467 (1955).

Berg, L., “Selecting the Agent for Distillation Processes,” Chem. Engr. Progr., 65 (9), 52 (Sept. 1969).

Biegler, L. T., I. E. Grossmann and A. W. Westerberg, Systematic Methods of Chemical Process Design, Prentice-Hall PTR, Upper Saddle River, New Jersey, 1997.

Black, C., “Distillation Modeling of Ethanol Recovery and Dehydration Processes for Ethanol and Gasohol,” Chem. Engr. Progr., 76 (9), 78 (Sept. 1980).

Chambers, J. M., “Extractive Distillation, Design and Application,” Chem. Engr. Progr., 47, 555 (1951).

Doherty, M. and M. Malone, Conceptual Design of Distillation Systems, McGraw-Hill, New York, 2001.

Drew, J. W., “Solvent Recovery,” in P. A. Schweitzer (Ed.), Handbook of Separation Techniques for Chemical Engineers, 3^rd ed., McGraw-Hill, New York, 1997, Section 1.6.

Ellerbe, R. W., “Steam Distillation/Stripping,” in P. A. Schweitzer (Ed.), Handbook of Separation Techniques for Chemical Engineers, 3^rd ed., McGraw-Hill, New York, 1997, Section 1.4.

Frank, T. C., “Break Azeotropes with Pressure-Sensitive Distillation,” Chem. Engr. Prog., 93 (4) 52 (April 1997).

Furter, W. F., “Production of Fuel-Grade Ethanol by Extractive Distillation Employing the Salt Effect,” Separ. Purific. Methods, 22, 1 (1993).

Hengstebeck, R. J., Distillation: Principles and Design, Reinhold, New York, 1961.

Hoffman, E. J., Azeotropic and Extractive Distillation, Interscience, New York, 1964.

Knickle, H. N., “Extractive Distillation,” AIChE Modular Instruction, Series B: Stagewise and Mass Transfer Operations. Vol. 2: Multicomponent Distillation, AIChE, New York, 1981, pp. 50-59.

Ludwig, E. E., Applied Process Design, 3^rd ed., Vol. 2, Gulf Publishing Co., Houston, 1997.

Luyben, W. L., “Azeotropic Tower Design by Graph,” Hydrocarbon Processing, 52 (1), 109 (Jan. 1973).

Nelson, P. A., “Countercurrent Equilibrium Stage Separation with Reaction,” AIChE J., 17, 1043 (1971).

Parkinson, G., “Distillation: New Wrinkles for an Age-Old Technology,” Chem. Engr. Prog., 101 (7) 10 (July 2005).

Perry, R. H. and D. W. Green (Eds.), Perry’s Chemical Engineers’ Handbook, 7^th ed., McGraw-Hill, New York, 1997.

Poffenberger, N., L. H. Horsley, H. S. Nutting and E. C. Britton, “Separation of Butadiene by Azeotropic Distillation with Ammonia,” Trans. Amer. Inst. Chem. Eng., 42, 815 (1946).

Prokopakis, G. J. and W. D. Seider, “Dynamic Simulation of Azeotropic Distillation Towers,” AIChE J., 29, 1017 (1983).

Robinson, C. S. and E. R. Gilliland, Elements of Fractional Distillation, 4^th ed., McGraw-Hill, New York, 1950, chap. 10.

Seader, J. D., “Distillation,” in Perry, R. H. and D. W. Green (Eds.), Perry’s Chemical Engineers’ Handbook, 6^th ed., McGraw-Hill, New York, 1984, Section 13.

Shinskey, F. G., Distillation Control, For Productivity and Energy Conservation, 2^nd ed., McGraw-Hill, New York, 1984, chaps. 9 and 10.

Siirola, J. J. and S. D. Barnicki, “Enhanced Distillation,” in Perry, R. H. and D. W. Green (Eds.), Perry’s Chemical Engineers’ Handbook, 7^th ed., McGraw-Hill, New York, pp. 13-54 to 13-85, 1997.

Smith, B. D., Design of Equilibrium Stage Processes, McGraw-Hill, New York, 1963, chap. 11.

Suzuki, I., H. Yagi, H. Komatsu and M. Hirata, “Calculation of Multicomponent Distillation Accompanied by a Chemical Reaction,” J. Chem. Eng. Japan, 4, 26 (1971).

Van Winkle, M., Distillation, McGraw-Hill, New York, 1967.

Wankat, P. C., “Graphical Multicomponent Distillation,” AIChE Modular Instruction, Series B: Stagewise and Mass Transfer Operations. Vol. 2: Multicomponent Distillation, AIChE, New York, 1981, p. 13-21.

Widago, S. and W. D. Seider, “Azeotropic Distillation,” AIChE Journal, 42, 96 (1996).

Woodland, L. R., “Steam Distillation,” in D. J. DeRenzo (Ed.), Unit Operations for Treatment of Hazardous Industrial Wastes, Noyes Data Corp., Park Ridge, New Jersey, 1978, pp. 849-868.

Woods, D. R., Process Design and Engineering Practice, Prentice Hall PTR, Englewood Cliffs, New Jersey, 1995.

Homework

A . Discussion Problems

A1. Compare the systems shown in Figures 8-1A, B, and C. What are the advantages and disadvantages of each system?

A2. Explain the differences between extractive and azeotropic distillation. What are the advantages and disadvantages of each procedure?

A3. Why is a cooler required in Figure 8-13? Can this energy be reused in the process?

A4. Explain the purpose of the liquid-liquid settler in Figures 8-3A, 8-4, and 8-18.

A5. Explain why the external mass balances are the same for Figures 8-3A and 8-6.

A6. Why are makeup solvent additions shown in Figures 8-13, 8-17, and 8-18?

A7. Explain why the pentane composition profile shows a maximum in Figure 8-19.

A8. Explain in your own words the advantages of doing reaction and distillation simultaneously.

A9. When doing distillation with reaction, the column should be designed both as a reactor and as a distillation column. In what ways might these columns differ from normal distillation columns?

A10. Reactions are usually not desirable in distillation columns. If there is a reaction occurring, what can be done to minimize it?

A11. Develop your key relations chart for this chapter.

A12. If a liquid mixture of n-butanol and water that is 20 mole % n-butanol is vaporized, what is the vapor composition? (see Figure 8-2). Repeat for mixtures that are 10, 30, and 40 mole % n-butanol. Explain what is happening.

A13. We plan to use extractive distillation (Figure 8-13) to separate ethanol from water with ethylene glycol as the solvent. The A product will be ethanol and the B product water. Stages are counted with condenser = 1, and reboiler = N. Feed plate for solvent recycle in column 1 is = NS. Feed stages are NF1 in column 1 and NF2 in column 2. Pick the best solution of the listed items.

1. If too much solvent is found in the water product,

a . Increase L/D in column 1.

b . Increase the value of NS.

c . Increase the number of stages in the stripping section of column 1.

d . Increase the number of stages in the stripping section of column 2.

e . Increase the number of stages in the enriching section of column 2.

2. If excessive water is found in the ethanol product when fresh solvent is added (with no solvent recycle),

a . Increase L/D in column 1.

b . Increase the solvent rate.

c . Increase the value of NS.

d . Increase the number of stages in the stripping section of column 1.

3. If excessive water is found in the ethanol product when solvent recycle is used, but was not when fresh solvent was used,

a . Increase L/D in column 1.

b . Increase the value of NS.

c . Increase the number of stages in the stripping section of column 1.

d . Reduce the water in the bottoms of column 2.

4. If excessive ethanol is found in the water product,

a . Increase L/D in column 1.

b . Increase the value of NS.

c . Increase the number of stages in the stripping section of column 1.

d . Reduce the water in the bottoms of column 2.

e . Increase the number of stages in the enriching section of column 2.

5. If too much solvent occurs in the ethanol product,

a . Increase L/D in column 1.

b . Increase the value of NF1.

c . Increase the number of stages in the stripping section of column 1.

d . Reduce the water in the bottoms of column 2.

C . Derivations

C1. Derive Eq. (8-7) for the two-column, binary, heterogeneous azeotrope system.

C2. For a binary heterogeneous azeotrope, draw the column arrangement if the feed composition is less than the azeotrope concentration (z < x_α). Show the McCabe-Thiele diagram for this system.

C3. For a binary heterogeneous azeotrope separation, the feed can be introduced into the liquid-liquid separator. In this case two stripping columns are used.

a . Sketch the column arrangement.

b . Draw the McCabe-Thiele diagram for this system.

c . Compare this system to the system in Figure 8-3A and Problem 8-C2.

C4. Sketch the McCabe-Thiele diagram for a two-pressure system similar to that of Figure 8-6.

C5. An equation for α_{org−w in w} similar to Eq. (8-13) is easy to derive; do it. Compare the predicted equilibrium in water with the butanol-water equilibrium data given in Problem 8-D2. Comment on the fit. Vapor pressure data are in Perry and Green (1984). Use the data in problem 8-D2 for solubility data.

C6. Derive Eq. (8-23).

D . Problems

*Answers to problems with an asterisk are at the back of the book.

D1. A distillation column is separating isopropanol and water at 101.3 kPa. We need x_p = 0.96 mole frac isopropanol, which the distillation column cannot produce. However, if the column gives y₁ = 0.80, a membrane separator can produce a product with x_p = 0.96. Some of this is returned as a saturated liquid reflux (x₀ = 0.96). The initial feed to the column is F = 1000 kg moles/hr and is 20 mole % isopropanol. This feed is a saturated vapor. The recycle steam is a saturated liquid and has a mole frac = 0.4. We desire x_B = 0.01. The column uses open steam, which is a saturated vapor and is pure water. We set the internal reflux ratio in the top section, L₀/V₁ = 5/9. Find the two optimum feed plate locations and the total number of stages. You may assume constant molal overflow (CMO). The column is similar to Figure 8-1C except that open steam heating is used. Isopropanol water equilibrium data are in Perry and Green (1984, p. 13-13). Isopropanol values are listed below.

D2.* VLE data for water-n-butanol are given in Table 8-2. We wish to distill 5000 kg moles/hr of a mixture that is 28 mole % water and 30% vapor in a two-column azeotropic distillation system. A butanol phase that contains 0.04 mole frac water and a water phase that is 0.995 mole frac water are desired. Pressure is 101.3 kPa. Reflux is a saturated liquid. Use L/V = 1.23(L/V)_min in the column producing almost pure butanol. Both columns have partial reboilers. ()₂ = 0.132 in the column producing water.

Table 8-2 Vapor-liquid equilibrium data for water and n-butanol at 1 atm. mole frac water

a . Find flow rates of the products.

b . Find the optimum feed location and number of stages in the columns.

Note: Draw two McCabe-Thiele diagrams.

D3. The VLE data for water and n-butanol is given in Table 8-2. (Note: this is mole % water.) We have flash distillation systems separating 100.0 kg moles/hr of two different water and n-butanol mixtures.

a . The feed is 20 mole % water and the vapor product is 40 mole % water. Find L, x, V, and T_drum.

b . The feed is 99 mole % water and 30% of the feed is vaporized. Find L, x, V, y and T_drum.

D4. A mixture of two liquid phases containing an average 88.0 mole % water and 12.0 mole % n-butanol is allowed to settle into two liquid phases. What are the compositions and the flow rates of these two phases? VLE data is in Table 8-2.

D5. We have a saturated vapor feed that is 80.0 mole % water and 20.0 mole % butanol. Feed rate is 200.0 kmole/hr. This feed is condensed and sent to a liquid-liquid separator. The water layer is taken as the water product, W, and the butanol (top) layer is sent to a stripping column which has a partial reboiler. The bottoms from this stripping column is the butanol product, which should contain 4.0 mole % water. Equilibrium data are in Table 8-2. Find:

a . Flow rates W and B

b . If = 4.0, find the number of equilibrium stages (step off from bottom up).

c . Determine for this separation.

D6. We plan to separate n-butanol and water in a continuous stripping column. Equilibrium data are in Table 8-2. The column has a partial reboiler. The distillate vapor from the column is sent to a total condenser and to a liquid-liquid separator where it separates into two liquid phases. These two phases are withdrawn as two distillate products (there is no reflux). The feed to the column is 100.0 kg moles/hr of a mixture that is 48.0 mole % n-butanol. The feed is a saturated liquid. The bottoms product is 8.0% water. The boilup ratio is 2.0. Assume CMO.

a . Find the distillate mole frac and the number of stages required.

b . Find B and the flow rate of vapor distillate.

c . Find the flow rates of the two liquid distillate products.

D7. Aniline and water are partially miscible and form a heterogeneous azeotrope. At p = 778 mm Hg the azeotrope concentrations are: The vapor is 0.0364 mole frac aniline, the liquid aqueous phase contains 0.0148 mole frac aniline, and the liquid organic phase contains 0.628 mole frac aniline. Estimate the relative volatility of water with respect to aniline in the organic phase. Note: Be careful to use the mole fracs of the correct component.

D8.* We have a feed of 15,000 kg/hr of diisopropyl ether (C₆ H₁₄ 0) that contains 0.004 wt frac water. We want a diisopropyl ether product that contains 0.0004 wt frac water. Feed is a saturated liquid. Use the system shown in Figure 8-4, operating at 101.3 kPa. Use L/D = 1.5 (L/D)_min. Determine (L/D)_min, L/D, optimum feed stage, and total number of stages required. Assume that CMO is valid. The following data for the diisopropyl ether—water azeotrope is given (Trans. AIChE, 36, 593, 1940):

y = 0.959, Separator: Top layer x = 0.994; bottom layer x = 0.012; at 101.3 kPa and 62.2°C. All compositions are weight fractions of diisopropyl ether.

Estimate α_{w−ether in ether} from these data (in mole frac units). Assume that this relative volatility is constant.

D9. We are using an enriching column to dry 100.0 kmole/hr of diisopropyl ether that contains 0.02 mole frac water. This feed enters as a saturated vapor. A heterogeneous azeotrope is formed. After condensation in a total condenser and separation of the two liquid layers, the water layer is withdrawn as product and the diisopropyl ether layer is returned as reflux. We operate at an external reflux ratio that is 2.0 times the minimum external reflux ratio. Operation is at 1.0 atm. Find the minimum external reflux ratio, the actual L/D, the distillate flow rate and mole frac water, the bottoms flow rate and mole frac water, and the number of equilibrium stages required (number the top stage as number 1). Use an expanded McCabe-Thiele diagram to determine the number of stages. Data for the azeotropic composition (Problem 8.D8) can be used to find the mole fracs of water in the two layers in the separator and the relative volatility of water with respect to ether at low water concentrations. The weight fractions have to be converted to mole fracs first.

D10.* A single-stage steam distillation system is recovering n-decane from a small amount of nonvolatile organics. Pressure is 760 mmHg. If the still is operated with liquid water present and the organic layer in the still is 99 mole % n-decane, determine:

a . The still temperature

b . The moles of water vaporized per mole of n-decane vaporized.

Decane vapor pressure is in Example 8-2. Water vapor pressures are (T in°C and VP in mmHg) (Perry and Green, 1997)

D11. We are doing a single-stage, continuous steam distillation of 1-octanol. The unit operates at 760 mm Hg. The steam distillation is operated with liquid water present. The distillate vapor is condensed and two immiscible liquid layers form. The entering organic stream is 90.0 mole % octanol and the rest is nonvolatile compounds. Flow rate of feed is 1.0 kg mole/hr. We desire to recover 95% of the octanol.

Vapor pressure data for water is given in problem 8.D10. This data can be fit to an Antoine equation form with C = 273.16. The vapor pressure of 1-octanol is predicted by the Antoine equation:

Log₁₀ (VP) = A − B/(T+C)

With A = 6.8379 B=1310.62 C=136.05

Where VP is vapor pressure in mm Hg, and T is in Celsius.

a . Find the operating temperature of the still.

b . Find the octanol mole frac in the vapor leaving the still.

c . Find the moles of octanol recovered and the moles of water condensed in the distillate product.

D12. Generation of distillation curves for systems with constant relative volatility is fairly straightforward and is an excellent learning experience. Generate distillation curves in Figure 8-7 for the benzene (A), toluene (B), cumene (C) system with α_AB = 2.4, α_BB = 1.0 and α_CB = 0.21. Operation is at total reflux.

a . *Mole frac of A in the reboiler is 0.006735 and mole frac of B is 0.003469. *

b . * Mole frac of A in the reboiler is 0.001 and mole frac of B is 0.009. *

c . Mole frac of A in the reboiler is 0.0003 and mole frac of B is 0.0097.

Remember the sum of the mole fracs of A, B, and C in the reboiler is 1.0.

*Solution is shown in Figure 8-7.

D13. Generation of residue curves for systems with constant relative volatility is fairly straightforward and is an excellent learning experience. Generate residue curves for the benzene (A), toluene (B), cumene (C) system with α_AB = 2.4, α_BB = 1.0 and α_CB = 0.21.

a. Mole frac of A in the stillpot is 0.9905 and mole frac of B is 0.0085. Compare this result to the distillation curves in Figure 8-7.

b. Mole frac of A in the stillpot is 0.99 and mole frac of B is 0.001.

Remember the sum of the mole fracs of A, B, and C in the stillpot is 1.0.

D14. We plan to use a two-pressure system (similar to Figure 8-6) to separate a feed that is 15.0 mole % benzene and 85.0 mole % ethanol. The feed rate is 100.0 kg moles/hr, and is a saturated liquid. The two columns will be at 101.3 kPa and 1333 kPa. We desire an ethanol product that is 99.2 mole % ethanol and a benzene product that is 99.4 mole % benzene. Assume the two distillate products are at the azeotrope concentrations.

a. Draw the flowsheet for this process. Label streams to correspond to your calculations.

b. Find the flow rates of the ethanol and benzene products.

c. Find the flow rates of the two distillate streams that are fed to the other column.

Data (Seader, 1984): Azeotrope is homogeneous. At 101.3 kPa the azeotrope temperature is 67.9°C and is 0.449 mole frac ethanol. At 1333 kPa the azeotrope temperature is 159°C and is 0.75 mole frac ethanol. Ethanol is more volatile at mole fracs below the azeotrope mole frac.

D15. We wish to separate a feed that is 60.0 mole % ethanol and 40.0 mole % benzene using a two pressure distillation system similar to Figure 8-6. The feed rate is 150.0 kmole/hr. The two columns operate at 101.3 kPa and 1333 kPa. Data on the azeotropes are given in Problem 8.D14. We want an ethanol product P_E that is 99.3 mole % ethanol and a benzene product P_B that is 99.5 mole % benzene. Assume that the distillate products approach within 0.01 mole frac of the azeotropic mole fracs (e.g., x_D from column at 1333 kPa is either 0.74 or 0.76).

a . Sketch the process flowsheet (label pressure and products) and the approximate McCabe-Thiele diagrams.

b . Determine ethanol product flow rate P_E and benzene product flow rate P_B.

c . Determine both distillate flow rates.

D16. An extractive distillation system is separating ethanol from water using ethylene glycol as the solvent. The makeup solvent stream is pure ethylene glycol. The flowsheet is shown in Figure 8-13. The feed flow rate is 100.0 kg moles/hr. The feed is 0.810 mole frac ethanol and 0.190 mole frac water. We desire the ethanol product (distillate from column 1) to be 0.9970 mole frac ethanol, 0.0002 mole frac ethylene glycol, and remainder water. The water product (distillate from column 2) should be 0.9990 mole frac water, 0.00035 mole frac ethylene glycol, and remainder ethanol.

Find flow rates of makeup solvent, distillate from column 1, and distillate from column 2.

D17. For the extractive distillation system shown in Figure 8-13, the saturated liquid feed is 60.0 mole % n-heptane and 40.0 mole % toluene. Feed flow rate is 100.0 kmole/hr. The solvent is phenol at a solvent recycle rate of 120.0 kmole/hr. We desire a n-heptane product (A-product) that is 99.2 mole % n-heptane and a toluene product (B-product) that is 99.6 mole % toluene. As a first approximation, assume the phenol is perfectly separated and Make-up Solvent = 0.0 kmole/hr. Find flow rates of n-heptane product & toluene product.

D18. An extractive distillation system is separating ethanol from water using ethylene glycol as the solvent. The makeup solvent stream is pure ethylene glycol. The diagram is in Figure 8-13 with ethanol as the A product and water as the B product. The feed flow rate is 100.0 kg moles/hr. The feed is 0.2000 mole frac ethanol and 0.8000 mole frac water. We desire the ethanol product (distillate from column 1) to be 0.9970 mole frac ethanol, 0.0002 mole frac ethylene glycol, and remainder water. The water product (distillate from column 2) should be 0.9990 mole frac water, 0.00035 mole frac ethylene glycol, and remainder ethanol.

Find flow rates of Makeup solvent, A Product and B Product.

D19. We wish to use n-hexane as an entrainer to separate a feed that is 80.0 wt % ethanol and 20.0 wt % ethanol into ethanol and water. The system shown in Figure 8-18 will be used. The feed is 10,000.0 kg/hr and is a saturated liquid. The ethanol product is 99.999 wt % ethanol, 0.001 wt % hexane, and a trace of water. The water product is 99.998% water, 0.002 wt % ethanol and a trace of hexane. Do external mass balances and calculate the flow rates of make-up solvent (n-hexane), ethanol product, and water product. (Assume that trace = 0.)

Note: Watch your decimal points when using weight fractions and wt %.

D20. We wish to separate 1500.0 kg moles/hr of a mixture of MEK and water using the system shown in Figure 8-6. The feed to column 1 is 60.0 mole % water. This column is at 14.7 psia and produces an essentially pure water bottoms. The column 1 distillate is slightly above the azeotrope mole % of 35% water (x_D1 = 0.353 water). This distillate is fed to column 2, which is at 100.0 psia. Column 2 produces a bottoms which is essentially pure MEK and a distillate of x_D2 = 0.496 mole frac water which is slightly less than the azeotrope (50% water at 100 psia). Operation is at steady state.

Find B₁, B₂, D₁, D₂, and the total feed to column 1.

D21. A distillation column is separating water from n-butanol at 1 atmosphere pressure. Equilibrium data are in Table 8-2. The distillation system is similar to Figure 8-4 and has a partial reboiler, a total condenser and a liquid-liquid settler. The bottom layer from the settler (rich in water with x_w = 0.975) is taken as the distillate product. The top layer (x_w = 0.573) is returned to the column as a saturated liquid reflux. The feed is 40.0 mole % water, is a saturated vapor and flows at 500.0 kg moles/h. The bottoms is 0.04 mole frac water. Use a boilup ratio of = 0.5. Assume CMO is valid. Step off stages from the bottom up. Find the optimum feed stage location and the total number of equilibrium stages needed.

D22. We are separating a saturated vapor feed that is 40.0 mole% water and 60.0 mole % n-butanol in an enriching column. Feed rate is 100.0 kmoles/hr. The vapor from the top of the column is sent to a condenser and then to a liquid-liquid separator. The aqueous layer is withdrawn as the distillate product and the organic layer is returned as reflux. A bottoms composition that is 20.0 mole % water is desired. Equilibrium data are in Table 8-2. Find:

a . The value of the external reflux ratio, L/D.

b . The number of equilibrium stages required.

c . The mole frac of water in the vapor, y₁, leaving stage 1.

D23. We are separating a mixture of n-butanol and water. The feed rate is 10.0 kmoles/hr. Feed is a saturated liquid and is 44.0 mole % water. The feed is sent to a kettle-type reboiler (aka a stillpot). The stillpot is an equilibrium contact. The bottoms from the stillpot is withdrawn continuously and has 32.0 mole % water. The vapor from the still pot is sent to a total condenser and then to a liquid-liquid settler (aka decanter). The distillate is the water layer from the settler (97.5% water). The organic layer from the settler is refluxed to the stillpot (57.3% water). Operation is at 1.0 atm. The VLE data for water and n-butanol at 1.0 atm is given in Table 8-2.

a . Find D and B.

b . Find L (rate of reflux liquid) and V (vapor rate from still pot to condenser).

The easiest way to attack this problem is from first principles. That is use equilibrium in the still pot plus mass balances along with the known mole fracs of the liquid layers in equilibrium in the decanter.

E . More Complex Problems

E1. We wish to separate water from nitromethane (CH₃ NO₂). The first feed (to column 1) is 24 mole % water, is a saturated vapor, and flows at 1000 kg mole/day. The second feed (to column 2) is 96 mole % water, is a saturated liquid, and flows at 500 kg moles/day. A two-column system with a liquid-liquid separator will be used. The bottoms from column 1 is 2 mole % water; use (L/V)₁ = 0.887 in this column. The bottoms from column 2 is 99 mole % water; use in this column. Data are given in Table 8-3.

Table 8-3. Water-nitromethane equilibrium data at 1 atm

a . Find the number of stages and the optimum feed location in both columns.

b . Calculate L/V in column 2. Explain your result.

E2. The VLE data for water and n-butanol at 1.0 atm is given in Table 8-2. The fresh feed is 100.0 kmole/hr of a saturated liquid that is 30.0 mole % water. This fresh feed is mixed with the return line from the separator to form the total feed, F_T, which is fed to a stripping column that produces the butanol product (see figure). Assume this total feed is a saturated liquid. The butanol product is 2.0 mole % water. The water product is produced from another stripping column and is 99.5 mole % water. In the butanol column the boilup ratio is 1.90. In the water column the boilup ratio is 0.1143. Assume CMO. Both reboilers are partial reboilers. Both reflux streams are returned as saturated liquids. Operation is at 1.0 atm. Since this problem is challenging, this list coaches you through one solution approach.

a . Find the flow rates of the two products (use external mass balances).

b . Find vapor flow rate in butanol column, then liquid flow rate in butanol column, which = F_T.

c . Calculate z_T from mass balance at mixing tee for fresh feed and reflux.

d . Plot the bottom operating line and find value of vapor mole frac leaving the butanol column and the number of equilibrium stages required for this column.

e . Use an expanded McCabe-Thiele plot to find the vapor mole frac leaving the water column and the number of equilibrium stages required.

E3. We wish to produce pure water by boiling it with n-decane vapor (see figure). This is sort of reverse of steam distillation. Seawater is roughly 3.5 wt % salt which can be approximated as NaCl. The feed is 1000 kg of seawater/hr. The feed temperature is 30°C. The water is heated with pure saturated n-decane vapor at 760 mm Hg. Most of the n-decane vapor condenses while the remainder is carried overhead with the water vapor. Pressure is 760 mm Hg. We wish to recover 60 % of the water as condensate.

a . Find the approximate still temperature.

b . Find the moles of n-decane carried over in the vapor/hr.

Data: VP of n-decane: Example 8-2, VP of water: Problem 8.D10, MW_water = 18.016, MW_NaCl = 58.45, MW_C10 = 142.28, Salt is non-volatile. Water and n-decane are immiscible. NaCl dissolves only in water.

F . Problems Requiring Other Resources

F1. A single-stage steam distillation apparatus is to be used to recover n-nonane (C₉ H₂₀) from nonvolatile organics. Operation is at 1 atm (760 mmHg), and the still will operate with liquid water present. If the still bottoms are to contain 99 mole % nonane in the organic phase (the remainder is non-volatiles), determine:

a . The temperature of the distillation

b . The moles of water vaporized per mole of nonane vaporized

c . The moles of water condensed per mole of nonane vaporized for a liquid entering at the still temperature

d . Repeat parts a and b if still bottoms will contain 2.0 mole % nonane

Data are available in Perry and Green (1984).

F2. We wish to recover a gasoline component, n-nonane (C₉H₂₀), from a non-volatile mixture of oils, grease, and solids. This will be done in a steady-state, single-stage, steam distillation system operating with liquid water present and at a total pressure of 102.633 kPa. The feed is 95.0 mole % n-nonane and we desire to recover 90% of the n-nonane in the distillate. The feed enters at the temperature of the boiler and the feed rate is 10.0 kmole/hr. Find:

a . The bottoms mole frac n-nonane in the organic layer.

b . The still temperature.

c . The kg moles of nonane in the distillate.

d . The kg moles of water in the distillate.

e . The kg moles of water in the bottoms.

Assume water and n-nonane are completely immiscible. Vapor pressure data for water is given in Problem 8.D10. Use the DePriester chart or Raoult’s law to obtain K_c9 (T, p_tot). Then p_org = K_c9 x_c9,org p_tot. Latent heat values are available in Perry’s and similar sources.

G . Computer Problems

G1. Solve problem 8.D23 with a process simulator.

Note: this can be done several ways if one uses Aspen Plus:

a . Use RADFRAC with N = 2 (total condenser and partial reboiler) with feed to the reboiler. Recycle organic from decanter to the column as a second feed (on the reboiler). Set bottoms rate and L/D. Start with L/D of 0.1 and work your way down by factors of ten to 0.0001 (which is almost zero).

b . Solve as a trial & error problem with FLASH2, a HEATER and a DECANTER. Return the organic layer from the decanter (reflux) as a second feed to Flash 2. Vary V/F in the drum until the desired bottoms mole frac is achieved. c. Other methods will work.

Chapter 8 Appendix Simulation of Complex Distillation Systems

This appendix follows the instructions in the appendices to Chapters 2 and 6. Although the Aspen Plus simulator is referred to, other process simulators can be used. The three problems in this appendix all employ recycle streams in distillation columns. The procedures shown here to obtain convergence are all forms of stream tearing. Since these are not the only methods that will work, you are encouraged to experiment with other approaches. If problems persist while running the simulator, see Appendix: Aspen Plus Separations Troubleshooting Guide, which is after Chapter 17.

Lab 7. Part A. Two-pressure distillation for separating azeotropes. A modification of the arrangement of columns shown in Figure 8-6 (or the very similar arrangement where the feed is input into the higher pressure column 2) can be used to separate azeotropes if the azeotrope concentration shifts significantly when pressure is changed. You should develop a system that inputs D₂ and F on separate stages.

We want to separate a feed that is 60 mole % water and 40 mole % methyl ethyl ketone (MEK). Feed is a saturated liquid and is input into column 1 as a saturated liquid at 1.0 atm. Fresh feed rate to column 1 is 100 kmole/hr. Column 1 operates at 1.0 atm. Start with N = 10, both feeds on N_feed = 5, and L/D = 1.0 in column 1. Distillate is a saturated liquid. Use a total condenser and a kettle-type reboiler. Set convergence at 75 iterations.

Column 2 is at 100 psia, and has a total condenser, saturated liquid distillate, and a kettle-type reboiler. Start with N = 15 and feed at N_feed = 7, with L/D = 2.0. Set convergence at 75 iterations.

When you draw the flowchart, put a pump (select under “pressure changers”) on the distillate line going from column 1 that becomes the feed to column 2. The resulting flowchart should be similar to 8-A1.

Figure 8-A1. Aspen Plus screen shot for two-pressure distillation system

To get started, do external mass balances and calculate accurate values for the two bottoms flow rates assuming that the water and MEK products are both pure. Specify the bottoms flow rate of column 1 = 60, but do NOT specify bottoms in column 2. In column 2 start with D = 20 and increase D in steps, 30, 40, 50, 60, 70 without reinitializing. If you don’t step up, Aspen Plus will have errors. Note that B in column 2 = 40 every time even though you did not specify it (Why?).

Although not a perfect fit, the Wilson equation gives reasonable VLE data. Use analysis to look at the T-y,x and y-x plots. Feel free to look for better VLE packages.

Find reasonable operating conditions to give water with a mole frac in the range from 0.90 to 0.95 from the bottoms of column 1 and MEK (mole frac 0.90 to 0.95) from the bottom of column 2. After you are happy with the design, run one more time, to size the trays for both columns. Use sieve trays and default values for the other variables.

When done, look at the distillate flow rates. Why are they so large? Note that they make the column diameters and heat loads large for this very modest feed rate. These large recycle rates and hence large diameters and heat loads make this design economical only when the shift of the azeotrope with pressure is rather large. MEK and water is an example that is done commercially.

Part B. Binary distillation of systems with heterogeneous azeotropes. The purpose of this part is to design a system similar to Figure 8-3A for separating n-butanol and water. Open a new blank file. Under Setup, check Met and make the valid phases vapor-liquid-liquid. List n-butanol and water as the components. Finding a suitable VLE model for heterogeneous azeotropes is a challenge. NRTL-RK was the best model of the half dozen I tried. (Feel free to try other models.) NRTL-RK fits the vapor composition of the azeotrope and the liquid composition of the water phase quite well (see Table 8-2, in problem 8.D2.) but misses a bit on the composition of the butanol phase. After setting up the decanter, use Analysis to look at the T-y,x and y-x plots. Compare with the data. Remember to allow vapor-liquid-liquid as the phases in Analysis.

The first step is to determine how the liquid-liquid settler works. Aspen Plus calls this a decanter (listed under separators). Draw a decanter with one feed and two product streams (no distillation columns yet). The feed to the decanter should be a mixture of n-butanol and water that you expect will separate into two liquid phases (e.g., 60 mole % water). Use a saturated liquid (typical decanter feed in Figure 8-3A). Now run the decanter system. Write down which product is the water phase (very little butanol) and which is the organic phase (almost 50% water).

The problem we want to solve is to separate 100 kmole/hr of a saturated liquid feed at 1 atm pressure. The feed is 80 mole % butanol and 20 mole % water. A system similar to Figure 8-3A is to be used. The columns will operate at 1.0 atm pressure. Use external mass balances to determine the flow rates of the two bottoms products assuming that they are essentially pure butanol and pure water (< E -04 mole frac of the other component).

Draw the flow diagram. Use total condensers for both columns. Both distillate products should be taken off as liquids and then be connected to the feed to the decanter. The hydrocarbon layer from the decanter should be connected to the feed of the distillation column producing pure butanol. The water layer from the decanter should be connected to the feed of the distillation column that produces pure water. Note that the Aspen Plus columns in 8-A2 do not look exactly like the arrangement in Figure 8-3A. We are using the condensers on the columns to condense the liquids. We will use the specification sheets to make the system behave like Figure 8-3A.

Figure 8-A2. Distillation system for separation of heterogeneous azeotrope

On the specification sheets for both distillation columns the lines from the decanter should be input as feed on stage 2 (first actual stage in the column). Use kettle-type reboilers. Set the bottoms rate for the butanol column at the values calculated from the external mass balances. Initially, the column that receives the feed works with N = 20 and feed at 10. Initially set the stripping column with 10 stages. Set the boilup rate to 1.0 in the stripping column.

Set the reflux ratios for both columns to 0.5. We don’t really want to reflux from the condensers, but want to take our reflux from the decanter; however, if you try setting the reflux ratio to 0.0 Aspen Plus will not run. Reduce both reflux ratios in steps down to about 0.025. These ratios are small enough that the results will be very close to using only reflux from the decanter phases. If you don’t do this in steps, the condenser will dry up and the run will have errors. Pamper Aspen Plus, and make it happy!

After you have successfully run the simulation, note that both columns are producing purer products than required. This implies that fewer stages and/or a lower boilup ratio can be used. Try adjusting these to achieve the same separation at lower capital cost (reducing N). Also, look for the optimum feed location in the first column.

Lab 8. Extractive Distillation. This assignment is more prescriptive and involves less exploration than other labs since convergence is often a problem with extractive distillation. The basic algorithm (Figure 6-1) assumes that the concentration loop will have little effect on the other loops. Extractive distillation systems have very nonideal VLE and this assumption is often not true. As always with Aspen Plus, it may be possible to obtain convergence by starting with a set of conditions that converges and slowly changing the variable of interest (e.g., L/D) to approach the desired value. When you have a convergence problem, reinitialize RADFRAC and then return to a condition that converged previously.

Problem. Use extractive distillation to break the ethanol-water azeotrope. Use the two-column system shown in Figure 8-13. The solvent is ethylene glycol. Both columns operate at a pressure of 1.0 atmosphere. The feed to column 1 is 100 kmoles/hr. This feed is a saturated liquid. It is 72 mole% ethanol and 28 mole % water.

In Figure 8-13 the A product will be the ethanol product and the B product (distillate from column 2) will be water. We want the A product to be 0.9975 mole frac ethanol (this exceeds requirements for ethanol used in gasoline). Use an external reflux ratio L/D = 1.0. (Normally these would be optimized, but to save time leave it constant.) The reflux is returned as a saturated liquid. This set of conditions should remove sufficient ethylene glycol from the distillate to produce an ethanol of suitable purity (check to make sure that this happens). The bottoms product from column 1 should have less than 0.000009 mole frac ethanol. This number is low to increase the recovery of ethanol.

The distillate from column 2 should contain less than 0.0001 mole frac ethylene glycol. The bottoms product from column 2 should contain less than 0.0001 mole frac water since any water in this stream will probably end up in the ethanol product when solvent is recycled.

Steps 1 and 2. First design the two columns in RADFRAC as columns in series with no solvent recycle. Thus, all of the solvent used must be added as the make-up solvent. For this part use a makeup solvent that is close to the expected concentration of the recycle solvent. That is, the solvent stream should be 0.9999 mole frac ethylene glycol and 0.0001 mole frac water. Use a solvent temperature of 80°C to approximately match the temperature of the solvent feed stage NS = 5 in column 1. Solvent pressure is 1.0 atmosphere. Use total condensers and kettle reboilers.

In the convergence section of the input block set the number of iterations to 75 (for both columns). (Go to Data in the menu bar and click on Blocks. On the left hand side of the screen click on the + sign next to the block for the distillation column you want to increase the number of iterations for. Then find Convergence and click on the blue check mark for convergence. This gives a table. Increase the maximum number of iterations to 75. Higher values don’t help.)

Design column 1. The solvent is treated as a second feed. On the flow diagram, use the feed port to add a solvent feed to the column. Then specify its location in the Table of input conditions. Add solvent at stage NS = 5. Use a solvent rate of S = 52.0 kmole/hr. Start with N of column 1 = 50 and feed location NF1 = 25. Specify the distillate flow rate that will give you 100% purity of the ethanol product and no ethanol in the bottoms product (do external balances). Find the optimum feed stage and the lowest total number of stages that will do the desired separation. Start by keeping NF1 fixed while you reduce N (total number of stages in column 1) to just obtain the desired bottoms and distillate concentrations. Then find the optimum feed stage, and try reducing N more. Low values of the feed stage (e.g., NF1 = 10) will probably not converge. Thus, start with a high number for the feed stage location and reduce NF1 slowly.

Design column 2. The feed to column 2 is the bottoms from column 1. It is a saturated liquid at 1.0 atmosphere. Specify the bottoms flow rate that will give you the desired purity of distillate and bottoms (do a very accurate external balance). Find an approximate (L/D)_min (N2 = 50 and NF2 = 25 is sufficient; start with L/D = 1.0 and move down). Operate with L/D = 1.15 (L/D)_min. Then find the optimum value of the feed stage (start with N2 = 20 and NF2 = 10). Note that column 2 is quite simple and a low value of L/D works.

Eventually we will connect the solvent recycle loop from column 2 to column 1. But first include a heat exchanger to cool the solvent. If you don’t do this column 1 will not work after you connect the solvent recycle loop (why not?). (To put the heat exchanger in the solvent line, go to Heat Exchangers and put a HEATER—used as a cooler in this case—in your flowsheet. Then left click on the solvent stream and right click on Reconnect Destination. Then connect the solvent stream to the arrow on the heat exchanger. Hit the Next button. You will get a window for the heat exchanger. Use 80°C. Pressure is 1.0 atmosphere.) To check that the HEATER is hooked up properly, try a run without changing the solvent makeup flow rate.

Connect the line leaving the heater to column 1 (use reconnect destination and connect it to the feed port for column 1). In the table of input conditions list the solvent makeup stream and the recycle line at the same stage (stage number 5).

To calculate the makeup solvent flow rate that you eventually want to use, do an external mass balance around the entire system,

Solvent Makeup flow rate = Solvent out in A product + solvent out in B product.

The resulting Makeup flow rate will be extremely small since losses of solvent are small (after all, no one wants to drink ethylene glycol with their alcohol or their water). If you immediately use this value of solvent makeup as a feed to the system, Aspen Plus will not converge. Start with the value you were using previously and rapidly decrease it (say by factors of roughly 5 or 10). Until the solvent makeup stream is at the desired value for the external mass balance, the “extra” ethylene glycol will exit with the distillate (water product) from column 2. This occurs because Bottoms flow rate in column 2 is specified and the only place for the extra ethylene glycol to go is with the distillate from column 2. (This is why you must set the bottoms rate in column 2. If you set the distillate rate in column 2 there is no place for the “extra” ethylene glycol to go and the system will not converge.) Ignore the values of the distillate flow rate and distillate compositions from column 2 until you use the Makeup solvent rate that satisfies the overall balance. Once you are close to the correct flow for makeup, change it to pure ethylene glycol. The final appearance of the system should be similar to 8-A3.

Figure 8-A3. Aspen Plus screen shot of completed extractive distillation system

If for some reason (not recommended) you have to uncouple the recycle loop, remember to reset the makeup flow rate to the desired value.

If necessary, make minor adjustments in stages, flows or reflux ratios to achieve desired purities. (This usually will not be necessary.)

As a minimum, record the following: the mole fracs of the two products and the two bottom streams; the heat duties of the two condensers, the two reboilers and the heat exchanger; the flow rates of all the streams in the process; L/D in each column; N and feed locations in each column; and the temperatures of the reboilers and condensers.