Equilibrium Stage Processes

A. Introduction There are many separations operations used in chemical processing where the objective is to have two phases form in a unit, come to equilibrium, and leave the unit as two separate streams. These two phases may be formed from a single inlet stream, or, they may be formed by mixing two inlet streams. In either case, we refer to such a process as an equilibrium stage. An example of the former is a flash vaporization unit, where a single inlet stream is brought to conditions such that a liquid phase and a vapor phase will develop and approach equilibrium in a vessel, which is commonly called a flash drum. The vapor and liquid phases are taken off as separate exit streams. The purpose is to separate morevolatile components from less-volatile components. In an ideal flash unit, equilibrium will be achieved. In practice, this will not occur, however, if properly designed, a flash unit may produce streams that are close to equilibrium compositions. A mixer/settler unit is an example of a process where two inlet streams as separate phases are contacted so that components are exchanged between them, approaching equilibrium as two phases, and then separated as two separate exit streams. This type of unit can be used in liquid-liquid extraction, liquid-solid extraction, distillation, and other diffusional operations. To achieve better separations of components than can be achieved in a single stage, chemical engineers use distillation towers or extraction trains containing many individual stages where streams can be contacted and brought to near equilibrium conditions. Calculational methods for these multistage operations are always based on the assumption of equilibrium in every stage. Stage efficiencies are then used to relate real operations to the ideal calculations based on equilibrium. These efficiencies depend on the fluid dynamics, heat transfer, and mass transfer in the real processes, and methods used in the estimation of stage efficiencies is beyond the scope of this class. B. Processes with a Single Inlet Stream (Flash Vaporization) In these processes, a single feed stream enters a unit, where it is allowed to come to equilibrium, forming two phases, which are removed as an overhead product and a bottoms product. The feed stream may be a single phase, or may already be in two-phase flow before entering the unit. The exit streams can be any two phases that can be separated: vapor and liquid, liquid and liquid, liquid and solid, etc. We will focus on a flash vaporization unit, where a liquid (or sometimes mixed liquid and vapor) feed is separated into an overhead vapor product and a bottoms liquid product, the two product streams assumed to be in equilibrium with each other. This is usually accomplished by suddenly lowering the pressure on the feed stream just as it enters the

flash drum. In practice, this is done by use of a valve to throttle the inlet stream. A diagram is shown below. P Vapor Feed F, xiFs, TF, PF, HF Q Bottoms (Liquid) L, xiLs, TL, PL, HL T Overhead (Vapor) V, yiVs, TV, PV, HV

In the variable list for each stream, the first symbol represents the molar flow rate, the second represents the mole fractions of the various components (some could be zero), and the last three represent temperature, pressure, and molar enthalpy. If there are n components in the system, and there are no chemical reactions, then there are n+3 variables for each of the three streams: 1 molar flow rate (F, V, or L), n-1 independent mole fractions (remember they add up to unity), and 3 other stream properties (temperature, pressure, and enthalpy). We have to include temperature and pressure, because the phase equilibrium relations depend on these variables, in addition to compositions of the phases. We include enthalpies, because they will appear in the energy balance. Q is the duty for the flash vaporization unit, that is, the heat rate term in the energy balance. It can be positive (energy added), negative (energy removed), or zero (adiabatic). The total number of variables is 3(n+3) + 3, where the last three are Q, T, and P: Total No. Variables = 3n+12 Now, lets write all of the equations that we can to provide relations between these variables. We can write materials balances, an energy balance, equilibrium relations, and enthalpy property relations. We will assume steady-state operation of the flash vaporization unit. The system will be the contents of the flash drum, with the throttling valve on the system boundary where the feed stream enters. Material balances: Total:

F =V + L
(i = 1 to n-1)

Components: FxiF = VyiV + LxiL

Energy balance: Q + FH F = VHV + LH L Note: The valve on the feed stream does not effect the energy balance, because the process of steady-state flow across a valve is considered an isenthalpic process. Equilibrium relations: Thermal: Mechanical: Species:
TV = TL = T

PV = PL = P
( T , P , y ' s) = f ( T , P , x ' s) (i = 1 to n) f iV iV iL iL

Note: To use the species equilibrium relations, we will have to provide specific equations, such as Raoults law, a modified Raoults law with activity coefficients, or fugacity relations from an equation of state. Enthalpy property relations:
H F = H F ( TF , PF , x iF ' s) HV = HV T , P , y iV ' s H L =H L
iL

( (T , P , x

) ' s)

Note: In the enthalpy relations, we have already noted that the temperatures and pressures of the exit streams are the same as the equilibrium conditions in the flash drum (from the equilibrium relations above). These relations are written in functional form, because they may come from a table or chart in a reference book (e.g., steam tables or an enthalpy-concentration diagram), or, more likely, they will be taken from some calculational scheme in a process simulation software package. The total number of equations we have written is 2n+8 [n (material balances) + 1 (energy balance) + n+4 (equilibrium relations) + 3 (enthalpy relations)]. Total No. Equations = 2n+8 To solve a real problem, the number of equations must equal the number of unknowns, thus, we can calculate the number of variables that must be specified to make the problem well-posed: No. Specifications = No. Var. - No. Eqns. = (3n+12) - (2n+8) = n+4 . Thus, we must specify n+4 variables before we can solve the above equations for the remaining unknown variables.

The most common set of variables to specify is as follows: F, xiFs, TF, PF, P, and (either Q or T) Note: An adiabatic specification for a flash unit means Q = 0, whereas, an isothermal specification means T = TF. Example Type I Problem. Flash Vaporization for a Benzene + Toluene Stream Problem: A stream containing 35.0 mol % benzene and 65.0 mol % toluene is flowing at 100.0 mol/s, 100.0 oC, and 10 atm. It is flashed across a valve and into a flash drum held at 100.0 oC and 1.00 atm. What are the compositions and flow rates of the product streams, and what is the duty for this flash unit? Prepare solutions by a) hand calculations assuming Raoults law, b) a graphical method using an enthalpyconcentration diagram, and c) PRO/II simulation. a) analytical solution using Raoults law Since this is a binary system, the specifications of T and P in the flash drum are sufficient to allow direct solution of the equilibrium relations. Remember that Raoults law assumptions say that the fugacity in the vapor phase can be replaced by the partial pressure (Py), and that the fugacity in the liquid phase can be taken as the vapor pressure times the liquid mole fraction (PSx):
Py1V = P1S x1 L Py 2V = P2S x 2 L

To use these equations, we must find vapor pressures for each component at temperature T = 100.0 oC. Using the Antoine equation parameters from p. 464 of . atm P2S = 0.732 atm Smith et al., we find: P1S = 1777 Replacing y2 by 1-y1, and combining the two equilibrium relations, the following solution is obtained:
y1V = 0.456 x1 L = 0.257 y 2V = 0.544 x 2 L = 0.743

Now, the material balance equations can be solved,

100.0 =V + L 100.0( 0.350) =V ( 0.456) + L ( 0.257) V = 46.7 mol / s L =53.3 mol / s

To solve the energy balance for the duty, we need to supply enthalpies for the streams. Because these components are very similar, a good assumption is that heats of mixing are small for both liquid and vapor mixtures:
H E =0 or H = xi H i

Thus, component liquid and vapor enthalpies are needed, along with known mole fractions, to calculate the mixture enthalpies. These could be taken from reference books, or estimated as follows. Choose as a reference H = 0 for the pure liquids at 100.0 oC. This will make both HF and HL equal to zero for our problem, because pressure has little effect on liquid-phase enthalpies. The component vapor enthalpies will then be closely approximated by the heats of vaporization at 100.0 oC, which can be estimated by the Watson method (p. 128 of Smith et al., 5th Ed., using boiling and critical temperatures from p. 635 of the same source, and using heats of vaporization at the normal boiling points from Table B.1 in Felder and Rousseau, 2nd Ed.):
1 0.663 H1V = 30765 1 0.627 1 0.628 H 2V = 33470 1 0.646
0.38

= 29600 J / mol
0.38

= 34100 J / mol

HV = 0.456( 29600) + 0.544 (34100) = 3.20 x10 4 J / mol Q = VHV + LH L FH F = 46.7(3.20 x10 4 ) + 0 0 = 1.49 x10 6 J / s

b) graphical solution using an H-x diagram Raoults law, or more complicated equilibrium relations, can be used to construct a T-x-y diagram for benzene + toluene at 1 atm. Material balances can be solved on this diagram. The compositions and flow rates of the vapor and liquid products can be found, but there are no enthalpies, so we could not find the duty.

1 0.9 120 0.8 0.7 0.6

T/C
110

T-x-y for Benzene + Toluene at 1 atm

Vapor Dew Points Liquid

0.5 0.4 0.3 0.2 0.1

100

90

Feed
0 0.2 0.4

Bubble Points
0.6 0.8 1

0 80

x,y

The above diagram was constructed using Raoults law. These calculations can be programmed in Excel using Antoine equations for the vapor pressures. Another easy way to obtain such a plot is to use the vapor-liquid equilibrium calculation on PRO/II, having selected the Ideal thermodynamics package. The feed line is drawn at the feed composition (x1F = 0.350). Where this intersects the 100.0 oC horizontal line (a tie line), the liquid product is to the left on the bubble-point curve, and the vapor product is to the right on the dew-point curve. As can be seen, x1L is about 0.25 and y1V is about 0.45, in agreement with our analytical solution. The amounts of vapor and liquid can be found using the lever rule. Thus, by visual inspection, the amounts of vapor and liquid product are about equal (50 mol/s each), which agrees roughly with the analytical results. A larger, moreaccurately-drawn graph would yield a better solution. A complete graphical solution can be obtained by use of a H-x-y diagram, but first we must construct one for this system. Three steps are needed to construct such a diagram: 1) location of the enthalpy line for vapors at their dew points, 2) location of the enthalpy line for liquids at their bubble points, and 3) location of tie lines connecting equilibrium liquid and vapor compositions at various temperatures. Although we need only the location of a single tie line at 100.0 oC to solve the current problem, we will construct the entire diagram for clarification and for future reference when multi-stage processes are discussed. We will retain the same reference conditions for enthalpy: HiL = 0 for both components at 100.0 oC. Also, we will continue to assume that the heats of mixing for liquid mixtures are small. Thus, the enthalpy of the liquid mixture which has a bubble point at 100.0 oC will have HL = 0 (see the analytical solution above). We can calculate the enthalpy for other bubble-point mixtures by

selecting a temperature (T), reading the liquid composition from the bubble-point curve on the T-x-y diagram (or calculating the equilibrium compositions directly from Raoults law), and performing the calculation indicated below:
H L = x1 L H 1 L + x 2 L H 2 L = x1 L

100

(C )

P 1L

dT + x 2 L

100

(C )

P 2L

dT

Heat capacity equations for the pure liquids can be found in Table C.3, p.639, of Smith et al., 5th Ed. The results for three points are shown below, which are sufficient to define the bubble-point enthalpy curve: at x1 = 0 at x1 = 0.257 at x1 = 1 HL = 1,800 J/mol HL = 0 HL = -3,000 (100.0 oC point)

Vapor enthalpies are found be adding the heats of vaporization to the liquid enthalpies that have just been calculated. Since the pressure is 1.00 atm, the heats of vaporization for the two components (x1 = 1 and x1 = 0) are just the values at their normal boiling points (80.0 and 110.6 oC, respectively). The values are 30,765 J/mol for benzene and 33,470 J/mol for toluene (from Felder and Rousseau, 2nd Ed., Table B.1). Thus, including the point at 100.0 oC calculated in the analytical solution, three points on the dew-point curve are given below, which are sufficient to define the nearly straight line that results: at y1 = 0 HV = 1,800 + 33,470 = 35,300 J/mol at y1 = 0.456 HV = 32,000 (100.0 oC point) at y1 = 1 HV = -3,000 + 30,765 = 27,800 The dew- and bubble-point compositions for any tie line (at any desired temperature) can be found by solving Raoults law, or read from the T-x-y diagram. The completed H-x-y diagram is shown as a sketch below, with the graphical solution to the current problem indicated.

H/ J mol-1

40,000 HV 108 oC L 100 oC 20,000 Feed 92 oC V 0 HL Liquid 0 0.5 x1, y1 From a larger plot (not shown), a decent graphical solution can be obtained. Of course, the compositions of the equilibrium products had to be found from the equilibrium relations (Raoults law) to construct the graph in the first place. The vapor and liquid product flow rates come from the lever rule, using measured distances (in mm) on the large graph:
V =F 17.5 mm V = (100.0 mol / s) = 48 mol / s V +L (17.5 + 19.0) mm L = F V = 100 48 = 52 mol / s

Vapor

1.0

The duty is found from the energy balance, using the enthalpies from the chart:
Q = VHV + LH L FH F = 48(32,000) + 0 0 = 15 . x10 6 J / s

Again, note that the enthalpy of the feed is taken as zero at 100.0 oC and 10 atm. The results agree very well with the analytical solution. Why would a person use a graphical solution, when the analytical solution is so easy? The answer will become apparent when we consider multistage processes, where analytical calculations by hand become tedious. Before high-speed computers, chemical engineers relied heavily on graphical methods. Now, we use them only to illustrate the principles behind the computer calculations.

c) solution from a PRO/II simulation

Components selected: Thermodynamics package: Feed stream properties: Flash unit specifications: Results:

benzene, toluene Ideal T = 100 oC, P = 10 atm F = 100 mol/s, x1 = 0.35 T = 100 oC, P = 1 atm Duty = 1.51x106 J/s V = 47.1 mol/s y1V = 0.455 HV = 47,567 J/mol HF = 15,721 J/mol L = 52.9 mol/s x1L = 0.256 HL = 15,939 J/mol

These results are within about 1 % of the analytical solution given above. Note that the enthalpies are relative to different reference conditions, but the energy balance gives almost the same result for Q: Q = VHV + LHL - FHF = 47.1(47,567) + 52.9(15,939) - 100(15,721) = 1.51x106 J/s Note that the enthalpy for the feed is slightly different from the enthalpy for the liquid product in the PRO/II simulation. This is one reason for the small difference between this result and the analytical result (1.49x106 J/s). C. Processes with Two Inlet Streams (Distillation Tower Tray, Mixer/Settler) In these process, two streams (different phases) enter the unit, are mixed and allowed to come to equilibrium, and are then separated into two product streams (different phases). A flow chart for such a unit is shown below, where streams L0 and V2 are mixed in the unit Stage 1, brought to equilibrium, and separate streams V1 and L1 exit. P V1 Stage 1 L0 Q The unit name and the numbering of the streams are chosen as indicated, so that we can later treat countercurrent multistage processes without changing the notation. Although L (liquid) and V (vapor) are used to represent stream flow rates (as would be most appropriate in distillation, gas absorption, or stripping), the phases could equally well be two liquids (liquid-liquid extraction), or one a fluid and the other a solid (e.g., liquid-solid extraction or gas adsorption). L1 T V2

For each stream, if there are n components in the system, there will be a set of n+3 stream variables (e.g., L0, xi0(i = 1,,n-1), T0, P0, HL0 for the liquid stream entering Stage 1). Thus, the total number of variables for the entire unit will be given by Total No. of Variables = 4(n+3) + 3 = 4n + 15 where the last 3 variables counted represent the unit variables T, P, and Q. Now, the number of equations that can be written for this unit is the same as for Type 1 processes (n material balances, 1 energy balance, n+4 equilibrium relations, 3 enthalpy relations), except that now one additional enthalpy relation can be written for the specific enthalpy of the fourth stream: Total No. of Equations = 2n+9 Thus, the number of specifications to make a real problem solvable is given by No. of Specifications = No. Var. - No. Eqns. = (4n+15) - (2n+9) = 2n+6 The most common set of variables that would be specified when considering a single stage are given below. Common Specifications: L0, xi0s, T0, P0, V2, yi2s, T2, P2, P, and (Q or T) This does not have to be the list specified, as any 2n+6 variables will suffice to allow a solution. However, some sets may be more-easily accommodated in process simulation packages. Example Type 2 Single-Stage Problem. Benzene + Toluene Equilibrium A liquid stream L0 contains 53.0 mol % benzene and the rest toluene. It is flowing at 38.0 kmol/h and 91.0 oC. This stream is mixed with a vapor stream V2 containing 26.0 mol % benzene, which is flowing at 62.0 kmol/h and 105.0 oC. The mixer/separator operates at 100.0 oC. The pressure is constant throughout this system at 1.00 atm. What are the flow rates and compositions of the product vapor and liquid streams? What is the duty for this mixer/separator? Compare analytical calculations, graphical results, and a PRO/II simulation.

a) analytical calculation

Lets first make a list of all the variables that are specified: L0 = 38.0 kmol/h V2 = 62.0 kmol/h T = 100.0 oC x10 = 0.530 T0 = 91.0 oC y12 = 0.260 T2 = 105.0 oC P0 = P2 = P = 1.00 atm

In this notation, ys are mole fractions in vapor streams, and xs are mole fractions in liquid streams. The first subscript refers to the component and the second subscript refers to the stream (numbered by the stage from which the stream originates). Thus, we have specified 2n+6 = 10 variables, and we should be able to solve for the rest. Once again, we have a binary system (benzene(1) + toluene(2)) which comes to phase equilibrium at 100.0 oC and 1.00 atm. From the Raoults law solution for the Example Type 1 Problem given above

y11 = 0.456

x11 = 0.257

Next, we must solve the material balance equations Total: Component 1:

L0 + V2 = L1 + V1

L0 x10 + V2 y12 = L1 x11 + V1 y11

Substituting the know values in these equations and solving gives

L1 = 46.9 kmol / h

V1 = 531 . kmol / h

Finally, the duty comes from an energy balance

Q + L0 H L 0 + V2 HV 2 = L1 H L1 + V1 HV 1

To solve this for Q requires us to estimate the four stream enthalpies. Again, taking the references to be HL = 0 at 100.0 oC for both components, and assuming negligible heats of mixing, the enthalpy of stream L1 will be zero. With these references, the enthalpies for all four streams can be read from the H-x-y chart constructed earlier (assuming pressure effects are small):
H L1 = 0 H L 0 = 1,500 kJ / kmol HV 1 = 32,000 HV 2 = 33,500

Then, the duty becomes

Q = 0.32 x10 6 kJ / h

b) graphical solution First, the feed points are located on the H-x-y diagram, and they are found to lie on the saturated liquid and vapor curves. If the vapor was significantly superheated, or the liquid greatly subcooled, then we would need enthalpy lines on the graph for these regions to be able to locate the feed points. The method of solution presented below would be exactly the same. The flow rates (L0 = 38.0 kmol/h and V2 = 62.0 kmol/h) specify a feed mixture point (by the lever rule), which is at a flow rate of 100.0 kmol/h, a composition of 0.36, a temperature of 101 oC, and an enthalpy of 20,100 kJ/kmol. But, the actual mixture in the unit must be at 100.0 oC, so, coming straight down from the feed mixture point to the 100.0 oC tie line, the actual mixture point in the unit can be found. By reading the difference in enthalpy between these two points, an enthalpy change of -3,200 kJ/kmol is required to bring a mixture of this composition to the proper temperature. Thus, the duty for this unit can be calculated from the total amount of mixture and the required change in enthalpy: Q = (100.0 kmol/h)(-3,200 kJ/kmol) = -0.32x106 kJ/h The end points for the 100.0 oC tie line tell us the compositions of the equilibrium streams leaving the unit: x11 = 0.26 y11 = 0.46

By use of the lever rule on the 100.0 oC tie line, both L1 and V1 can be found: L1 = 47 kmol/h c) PRO/II simulation results L1 = 46.6 kmol/h x11 = 0.256 V1 = 53.4 kmol/h y11 = 0.455 Q = -0.31x106 kJ/h V1 = 53 kmol/h

D. Countercurrent Multistage Operations (Extraction, Gas Absorption, Stripping)

To obtain good separations in processes such as liquid-liquid extraction, gas absorption, and stripping, multiple stages are often used in practice. One common arrangement is to have countercurrent flow through a series of stages. A simple flowchart for such a process with m stages is shown below. The advantage of this arrangement is that the driving force for mass transfer is more nearly constant throughout this system, the streams L0 and V1 being the richest in transferable component(s), and streams Lm and Vm+1 being the most dilute. V1 1 L0 L1 V2 2 L2 Lm-2 V3 Vm-1 m-1 Lm-1 Vm m Lm Vm+1

By counting variables and equations, we obtain No. Variables = (2m+2)(n+3) + 3m = 2mn+9m+2n+6 No. Equations = mn + m + m(n+4) + (2m+2) = 2mn+7m+2 No. Specs. Needed = 2n+2m+4 There are two common problems that the chemical engineer faces. In the first, the number of stages is known, and the separation is to be calculated (how the compositions of the exit streams compare with the entering streams). In the second, the degree of separation is specified, and the number of stages needed is to be calculated. The former might involve determining what separation could be achieved with existing equipment. The latter might involve design of new equipment to achieve a desired separation. For example, one possible set of specifications would be the following (assuming that the number of stages, m, is given): L0, xi0s, T0, P0, Vm+1, yim+1s, Tm+1, Pm+1, and Q, P for all m stages This is still a rather simple result. We need to specify the flow rates, compositions, and conditions (T and P) of the two feed streams to the overall process, plus two specifications for each stage. If Q = 0 is taken for each stage, then all of the stages are adiabatic, not an uncommon assumption in practice. An alternate assumption that is often employed is that all stages operate at the same temperature, which would be termed the isothermal assumption. If P is taken to be a constant throughout the entire process, then pressure drop through the stages is being neglected. The analytical solution is now more complicated than for a single stage, because, with the specifications indicated above, we do not know values for the flow rate, compositions, and conditions of two streams for any single stage, which would be needed to get the

calculations started. Note that we can not solve for the exit streams Lm and V1 by overall material balances, because we can not predict the combined effect of all the equilibria for the m stages. On the other hand, if we had specified streams L0 and V1, we could calculate stage 1, then proceed to stage 2, etc. When we reached stage m, we would know all unknown stream and unit variables. With the specifications indicated above, the analytical problem is trial and error, in that assumptions have to be made about stage 1 (e.g., specifying variables for stream V1) to allow stage-to-stage calculations to be made for stages 1, 2, 3, .., m. Then, a check has to be made, to see if the known properties of stream Vm+1 are duplicated. If not, then the assumptions for stage 1 need to be modified, and a new set of stage-to-stage calculations performed. This needs to be repeated until convergence is achieved (hopefully). Although modern computers can do these stage-to-stage calculations very rapidly, the convergence problem is a difficult one. This is because there are no simple, fool-proof ways to adjust the assumptions about stage 1, to converge on the correct answers for stream Vm+1 (at the other end of the process). Chemical engineers are often frustrated because it is so difficult to get computer simulation packages for multicomponent stagewise processes to converge. The Concepts of Net Flow and the Delta Point Consider material balances for a system including stages 1, 2, , j in the flowchart shown above. We can write a total material balance, n-1 component material balances, and an energy balance around this system: L0 + Vj+1 = Lj + V1 L0xi0 + Vj+1yij+1 = Ljxij + V1yi1 L0HL0 + Vj+1HVj+1 = LjHLj + V1HV1 These equations can be rearranged to give the equations of net flow between any two stages: = Lj - Vj+1 = L0 - V1 xi = Ljxij - Vj+1yij+1 = L0xi0 - V1yi1 H = LjHLj - Vj+1HVj+1 = L0HL0 - V1HV1 Since we are assuming steady-state operation, the flow rates and properties of streams L0 and V1 are not changing, thus, we can conclude from these net flow equations that the net flow to the right (of total mass, mass of component i, or energy) between any two stages is a constant, independent of where we are in the series of stages. Note that net flows may be positive or negative, depending on whether the L or V streams carry more material (or energy). We can define a fictitious concentration for component i, and a fictitious enthalpy, by combining the above net flow equations:

xi = (Ljxij - Vj+1yij+1)/(Lj - Vj+1) H = (LjHLj - Vj+1HVj+1)/( Lj - Vj+1) These equations may be used on phase diagrams to locate -points, which are constant for a series of stages, and provide the relationship between streams that are passing between stages. For example, the composition of component i for the -point (xi) lies on a straight line through x0 and y1, and also on a straight line through xm and ym+1, the composition points at the other end of the series of stages. Delta points (relating streams passing between stages) and tie lines (relating equilibrium streams leaving any stage) are all that need to be known to step from stage-to-stage on a phase diagram, as will be shown in the example below. Example Countercurrent Multistage Problem. Extraction of acetic acid from water with isopropyl ether. A three-stage mixer-settler system is to be used to extract acetic acid from 100. lb/h of a 50. wt % aqueous solution, with the solvent being 200. lb/h of pure isopropyl ether. The feed streams are at 68 oF and 1 atm, and the stages are all assumed to be isothermal and isobaric. Estimate the flow rates and concentrations of the extract and raffinate streams leaving this extraction train, using both analytical and graphical solutions. What percentage of the acetic acid is extracted?

Acetic Acid (1) + Water (2) + Isopropyl Ether (3)

V1 1 L0 = 100. lb/h x0: x10 = 0.50 x20 = 0.50 x30 = 0.00

V2 2 L1

V3 3 L2 L3

V4 = 200. lb/h y4: y14 = 0.00 y24 = 0.00 y34 = 1.00

Equilibrium data: Refer to the handout graph, which is a triangular phase diagram, roughly reproduced below. Compositions are in mass fractions. The heavy curve represents the equilibrium data. Tie lines are shown as dashed lines. These lines connect equilibrium points in the two partially miscible phases. Any stream flowing from a unit must lie on either the upper or lower part of this

equilibrium curve. We will use the other construction lines in the graphical solution below.

-point

Isopropyl Ether (3) 1 y1

y4 zF

x3 0 0 Water (2) a) analytical solution First, lets see if the problem is well-posed. There should be 2n+2m+4 = 16 specifications. We know six flow rates and compositions, plus the temperature and pressure for these two feed streams. In addition, we know that T = 68 oF and P = 1 atm for each stage. Thus, there are 16 specifications. Because we know the temperature and pressure in each stage, the energy balance equations do not have to be solved to find the equilibrium conditions for each stage. They could be used to find the Q for each stage, but this is not part of the problem statement, so we will ignore the energy balances. x0 1 Acetic Acid (1)

To start the calculations we can assume a flow rate and composition for stream V1 (remember that this point must lie on the upper part of the equilibrium curve): V1 = 250 lb/h, y1: y11 = 0.20, y21 = 0.05, y31 = 0.75

We can then solve the material balances for stage 1, using the tie lines on the handout phase diagram to determine x1 from the y1 point assumed: x1: x11 = 0.35, x21 = 0.60, x31 = 0.05 The three material balance equations can be written for stage 1: L0 + V2 = L1 + V1 L0x10 + V2y12 = L1x11 + V1y11 L0x20 + V2y22 = L1x21 + V1y21 Although it looks like there are four unknowns (V2, L1, y12, y22) , we must remember that the point y2 must also lie on the upper part of the equilibrium curve, because it is for a stream leaving stage 2. Thus, y12 and y22 must be related. From the phase diagram, this relation is nearly a straight line (up to y12 of about 0.25): y2j = 0.15y1j + 0.02 The analytical solution of the above equations for stage 1 then becomes L1 = 76.8 lb/h, V2 = 226.8 lb/h, y2: y12 = 0.119, y22 = 0.038, y32 = 0.843 Similarly, for stage 2, we use a tie line and then solve the balance equations: x2: x12 = 0.26, x22 = 0.71, x32 = 0.03 L2 = 62.2 lb/h, V3 = 212.2 lb/h, y3: y13 = 0.076, y23 = 0.031, y33 = 0.893 Finally, for stage 3 (with y14 = y24, since the point for V4 does not have to lie on the equilibrium curve), x3: x13 = 0.21, x23 = 0.76, x33 = 0.03 L3 = 68 lb/h, V4 = 218 lb/h, y4: y14 = y24 = 0.06, y34 = 0.88 Note that rounding has been done to 2 significant figures, because the phase diagram could only be read to this level of accuracy. These results do not quite agree with the known values for stream V4 (see the flowchart above), so we should choose different values for V1 and y11 and repeat the stage-to-stage calculations. Let the 2nd set of assumptions be V1 = 258 lb/h, y1: y11 = 0.18, y21 = 0.04, y31 = 0.78 By stage-to-stage calculations, this set of assumptions leads to the result

L3 = 51 lb/h, V4 = 209 lb/h, y4: y14 = 0.01 , y24 = 0.01 , y34 = 0.98 which are close enough to the desired results to assume convergence. Thus, the final solution is as follows: Feed: L0 = 100. lb/h x0: x10 = 0.50 x20 = 0.50 x30 = 0.00 V4 = 209 lb/h y4: y14 = 0.01 y24 = 0.01 y34 = 0.98 L3 = 51 lb/h x3: x13 = 0.13 x23 = 0.84 x33 = 0.03

Solvent:

Raffinate:

Extract:

V1 = 258 lb/h y1: y11 = 0.18 y21 = 0.04 y31 = 0.78

acetic acid extracted = (0.18x258 - 0.01x209)x100%/(0.50x100) = 89 % b) graphical solution Since this problem does not involve the energy balance, we do not have to use enthalpies, and the solution can be constructed on the triangular composition diagram, with mass fraction acetic acid on the horizontal axis and mass fraction isopropyl ether on the vertical axis. We will find this solution relatively easy to construct, as the equilibrium data are initially available in graphical form. The first step is to locate the feed and solvent stream points, and the mixture point representing the combination of these two streams. The feed point x0 is at 0.5,0 and the solvent point y4 is at 1,0 . A line is drawn between these two points, and the feed mixture point zF is located on this line by the lever rule (see above diagram): L0/V4 = 100/200 = 0.5 By an overall material balance for the entire extraction train, this zF mixture point must also lie on a line connecting the composition points for the two exit streams: x3 and y1. Assuming a position for the point y1 (on the upper phase boundary), immediately defines the point x3 (on the lower phase boundary). By the lever rule, the flow rates L3 and V1 can be found, since, by overall material balance, they must add up to 300. lb/h (the sum of the two feed streams).

For example, if y11 = 0.18 is assumed, then x13 = 0.10, L3 = 42 lb/h, and V1 = 258 lb/h are found from the diagram. Next, the -point is found by the intersection of two lines representing streams passing each other: the x0-y1 line and the x3-y4 line. This construction is shown on the figure above. The -point is the point that lets us graphically solve the material balances to relate streams passing each other between extraction units. The plates are stepped off by starting at point y1, following a tie line to x1 (on the lower phase boundary), then following a line toward the -point up to y2 (on the upper phase boundary), then following a tie line down to x2, then a -point line up to y3, then a tie line to x3. If x3 = 0.10, then we assumed the correct y1 starting point. In fact, the result is about x3 = 0.12, which is pretty close, probably as close as we can read this diagram. E. Feed Streams, Side Draws, Reflux, and Reboil (Distillation Towers) If we tipped the m stages in the above figure so they ran vertically down the page, it would almost look like a distillation tower with m trays inside. If fact, it could well be a tray-type gas absorber or steam stripper. The differences for a distillation tower would be that the feed is usually introduced somewhere in the middle, side product streams may be taken at various locations, and only one product stream comes off each end through the use of reflux and reboil. Below is a flowchart for a distillation process with one liquid feed stream F, one liquid overhead product stream D, and one liquid bottoms product stream B. Condenser V2 C QC D, xiD 1 L1 Lf-2 F, xiF Vf-1 f-1 f Lf Lm-2 QR B, xiB Vf+1 Vm-1 m-1 m Reboiler

Remember that the right side is really the bottom of the tower, and the left side is really the top. The condenser is pictured as a total condenser, and, as such, is not a stage, because there are not two streams leaving this unit in an equilibrium situation. We have not counted the condenser in the m stages shown. Note that some computer simulation packages may number this unit, even when it is not an equilibrium stage. In some situations, it may be

very difficult to condense all of the overhead vapors, and the condenser may be employed as a partial condenser. In that case, it should be numbered as a stage, and the product stream becomes a vapor stream, rather than a portion of the liquid condensed. The reboiler is pictured as a partial reboiler, and, as such, is an equilibrium stage. We have counted the reboiler in the m stages shown. By tradition, we will designate properties of the reboiler with a subscript R rather than m (QR, TR, and PR). The feed plate (plate number = f) is taken to be the plate just below the point at which the feed is introduced. In the flowchart, we have assumed the feed a liquid. In practice, the feed could equally well be a vapor, or a mixture of liquid and vapor, in which case it should join the Vf stream, or part into the Vf stream and part into the Lf-1 stream. This would not change the number of specifications that we shall determine below. The duties of the condenser (QC) and reboiler (QR) are shown. Often, a tower operating at temperatures well removed from the temperature of its environment will be well insulated, and all of the other stages will be assumed adiabatic, or Qj = 0 for j = 1,,m-1. Lets once again count variables and equations, so that we can define how many specifications are needed to form a well-posed distillation tower problem: No. Variables = (2m+5)(n+3) + 3(m+1) = 2mn + 9m + 5n + 18 No. Equations = (m+1)(n+1) + m(n+4) + 2 + (2m+5) + (2n+3) + (n+2) = 2mn + 7m + 4n + 13 No. Specifications = 2m + n + 5 The difficulty in preparing the above list is counting the equations, as the condenser equilibrium relations (2), splitter relations (2n+3) and mixing point relations (n+2) require special consideration to get the number of equations right. A typical list of specifications would be one of the following: 1) F, xiFs, TF, PF, PC, TC(exit), QC, QR, PR, and Qj, Pj (j=1,,m-1) 2) F, xiFs, TF, PF, PC, TC(exit), D, L0/D, PR, and Qj, Pj (j=1,,m-1) In the second list, L0/D is called the reflux ratio. It is assumed in our analysis that the number of components (n) and the number of stages (m) are known, however, as another option, it is possible to find the number of equilibrium stages needed to obtain a specified separation of a key component in the feed. Some simulation packages will allow this to be done directly. In others, the number of stages must be specified, and then changed by trial and error until the desired separation is obtained.

Stage-to-stage calculations for a distillation tower are often iterative, because there is often not enough information known initially for any stage to allow calculations to proceed from one end of the tower to the other. Instead, a trial and error approach must be taken, with initial assumptions made for some specified stage, followed by stage-tostage calculations through the tower, taking the feed into account at the appropriate point. If all the specifications are satisfied, the calculation is stopped. If not, adjustments are made, and the stage-to-stage calculations repeated. The Ponchon-Savarit Method for Binary Systems As is often the case in chemical processing, the analytical calculations are simplified for binary systems, and a graphical solution is possible on an enthalpy-concentration diagram. The only real difference from the countercurrent multistage solutions, given in the preceding section, is that the net flows are different in the sections of the tower above and below the feed location. Thus, two delta-points will be needed to allow stages to be stepped off throughout the tower. The upper section of a distillation tower is called the enriching (or rectifying) section, and the lower section is called the stripping section. In the upper section, the vapor stream is being enriched in the more-volatile component(s) by removal of the heavier component(s). In the lower section, the more-volatile component(s) is being stripped out of the liquid stream. The net flow equations become Enriching Section: Stripping Section: = L0 - V1 = Lj - Vj+1 = -D = Lm = Lj - Vj+1 = B

The coordinates of the two delta-points become Enriching Section: Stripping Section: x = xD, H = HD - QC/D = HD + [(L0/D)+1](HV1-HD) x = xB, H = HB -QR/B

Note: QC will be negative, so the delta-point in the enriching section will be above HD, the enthalpy of the overhead product stream. The second form for H is useful when the reflux ratio (L0/D) is given. QR will be positive, so the deltapoint in the stripping section will be below HB, the enthalpy of the bottoms product stream. Note: The two delta-points and the feed point zF must lie on a straight line, so, if one delta-point and the feed point are known, then the other delta-point can be located on the enthalpy-concentration diagram (as the intersection of this straight line with a vertical line at xB).

Once the delta-points have been located, the stages can be stepped off, using the upper delta-point and the tie lines in the enriching section. Then, once the feed point zF has been passed, the lower delta-point and the tie lines are used to step off stages in the stripping section. This Ponchon-Savarit method will be illustrated in the example problem below. The Approximate McCabe-Thiele Method A simplified procedure for binary distillation can be used under the assumption of constant molal overflow. What this means is that L and V remain constant throughout the rectifying section of a tower, and constant at different values (L and V) throughout the stripping section. For this to be valid, a number of assumptions have to hold: Assumptions: 1. The molar heats of vaporization of all mixtures are equal (distance between the bubble-point and dew-point curves on the H-x-y diagram everywhere the same). 2. The heats of mixing, heat losses, and sensible heat effects due to temperature gradients are all negligible. The assumption of constant molal overflow is not usually made in modern simulation packages. However, in many cases this is a reasonable approximation, and the McCabeThiele method is sometimes used as a quick hand calculation to give a quick, visual check of what is happening in the simulation. This assumption constant molal overflow allows temperatures and enthalpies to be ignored, and stages to be stepped off on a y-x diagram for the binary system at the pressure of the tower (assumed constant). The equilibrium curve on the y-x diagram can be found from experimental data, taken from a calculated T-y-x diagram, or calculated directly from the phase equilibrium relations. By tradition, the more volatile component is chosen as component 1, so that the y-x equilibrium is usually above the 45o line where y = x. Of course, if there is an azeotrope, then the equilibrium curve will cross the 45o line at the azeotrope composition. To finish the construction of this graphical solution, we have to locate operating lines on the y-x diagram. By material balances around the top end of the column, with L and V constant, an equation can be easily derived for the straight operating line for the enriching section y = Dx L x + D V V

where y and x are mole-fraction compositions for any two streams passing between stages.

Similarly, for the stripping section the operating line becomes

y=

Bx L x B V V

These two operating lines cross on the feed line, which is a vertical line at x = xF for a saturated liquid feed stream. For conditions other than saturated liquid for the feed stream, refer to Perrys Handbook or a Unit Operations text for details on construction of the feed line. For a saturated liquid feed at rate F, L = L + F and V = V

Since, by material balances, V = L + D, F=D+B and FxF = DxD + BxB

we can locate both operating lines if F, xF, D, xD, and L/D are known. We could know B, xB rather than D, xD and still make the construction. Starting at the top of the tower, where y1 = xD (on the 45o line) , we can find x1 by moving horizontally to the left to the equilibrium curve, then y2 by moving vertically down to the operating line, then x2 horizontally on the equilibrium curve, then y3 vertically on the operating line, etc. This procedure is followed until the feed line is passed, then the operating line for the stripping section is used to step off stages to the bottom of the tower. This McCabe-Thiele construction will be illustrated in the example problem below. Example Distillation Problem. Separation of Benzene and Toluene. An equimolar benzene (1) + toluene (2) liquid stream is flowing at 100.0 kmol/h, 1.00 atm, and its bubble-point temperature. It is to be separated into two liquid streams by use of a distillation tower. The overhead product stream is to be 50.0 kmol/h and contain at least 95 % of the benzene fed to the tower. A total condenser (with no subcooling of the liquid), a reflux ratio of 3.00, and a partial reboiler are to be used. What is the minimum number of trays required? What are the duties of the condenser and reboiler? Assume negligible pressure drop, and no heat losses from the column. Use both graphical solution methods indicated above and compare with a PRO/II simulation.

a) graphical solution (Ponchon-Savarit)

Given: D = 50.0 kmol/h Equilibrium and enthalpy data:

x1D = 0.95(0.50)(100.0)/50.0 = 0.95 H-x-y phase diagram (handout)

Location of -point: H = HD +(L0/D+1)(HV1-HD) = (-2900) + 4[28200-(-2900)] = 121500 kJ/kmol x = x1D = 0.95 Note: QC can be obtained from the alternate equation given for H: QC = D(HD-H) = 50.0(-2900-121500) = -6.22x106 kJ/h Location of feed point: zF = x1F = 0.50 HF = HL = -1200 kJ/kmol (bubble point liquid)

Location of -point: (straight line through -point and feed point zF to xB) B = F - D = 50.0 kmol/h xB = x1B = 0.05(0.50)(100.0)/50.0 = 0.05 H = -123900 kJ/kmol (from graph) Note: This point could also be obtained analytically using an energy balance around the entire tower to obtain QR, and then using the equation given above for H QR = DHD + BHB - FHF - QC = (50.0)(-2900)+(50.0)(1500)-(100.0)(-1200) -(-6.22x106) = 6.27x106 kJ/h H = HB-QR/B = 1500-(6.27x106)/50.0 = -123900 kJ/kmol Stepping off stages: Remembering that the composition of stream V1 is the same as stream L0, we start at y11, and follow down a tie line to x11 (the composition for stream L1), and then a line back toward the -point to y12, then down a tie line to x12, etc. This construction is continued until the feed point is passed (on tray 4), then the -point is used to relate passing streams between stages in the stripping section. Nine equilibrium stages are required to surpass a bottoms composition of 0.05 mole fraction. This is the minimum number required. If the stages have efficiencies less than 100 %, then the number of actual stages required will be greater than nine.

b) graphical solution (McCabe-Thiele) The first step is to construct a y-x diagram for the benzene + toluene system at 1 atm pressure. This can be obtained from the T-y-x diagram discussed above. Perhaps the easiest way to obtain a useful copy is to use the binary phase equilibrium calculation on PRO/II, which will generate the required y-x plot, which can then be printed directly. A handout has been prepared with the graphical solution to this example on it. The feed, overhead, and bottoms compositions are found (by material balances if necessary). In this problem, they are xF = 0.50, xD = 0.95, and xB = 0.05 (mole fractions of benzene). The feed line is drawn vertically at xF = 0.50, from the 45o line upward (for saturated liquid feed). Then the operating line for the enriching section is drawn from its upper end at y1 = x0 = xD (on the 45o line), with a slope L/V = (L/D)/(L/D+1) = 3.0/4.0 = 0.75, terminating on the feed line. The operating line for the stripping section is drawn from where the operating line for the enriching section terminates on the feed line, down to the bottoms point on the 45o line (at xB). Stages are stepped off from the top of the tower, between the operating line and equilibrium curve, switching over to the stripping section operation line once the feed line is passed. From this diagram, it is concluded that between 9 and 10 equilibrium trays are needed (including the reboiler). It should be noted that the optimum feed location would be on tray 5. If the feed is placed on a tray higher or lower in the tower, then more equilibrium trays will be needed to carry out the same separation. In these cases, one of the operating line would have to be extended, with the switch-over occurring above or below the feed line. A pinch point could be encountered, where the extended operating line would intersect the equilibrium curve, resulting in infinite trays being required. Care must be taken in practice to avoid pinch points in the operation of distillation towers. A McCabe-Thiele diagram will indicate if a near pinch-point condition exists in a tower, which may be difficult to detect from a simulation output. If a tower design is too near a pinch point, then it may be risky to try to implement that design.

c) PRO/II solution

A distillation tower was selected and placed on the flowchart. Based on the graphical solution, 10 trays were chosen for the computer simulation. Remember, PRO/II counts the condenser as a tray, even though it is a total condenser, and, thus, not an equilibrium stage. Thus, the simulation was for a tower with 8 trays, a partial reboiler (a 9th equilibrium stage), and a total condenser. A feed stream, on overhead liquid product stream, and a bottoms stream were placed on the flowchart. The components benzene (1) and toluene (2) were chosen, as was the idealsolution thermodynamic system IDEA01. The feed stream rate (100.0 kmol/h), composition (x1 = x2 = 0.50) , and conditions (P = 1.00 atm, bubble point) were set. For the tower, the feed was set on tray 5. For the condenser, the type total bubble point was selected, and the pressure set at 1.00 atm. A kettle-type reboiler was selected. The pressure profile was set at 1.00 atm at the top, with no pressure drop through the tower. Two tower specifications were written. First, the overhead rate was set at 50.0 kmol/h. Second, the reflux ratio was set at 3.00 on a mole basis. The specified variables for these two specifications were chosen to be QC and QR. Results: QC = -6.21x106 kJ/h QR = 6.28x106 D = 50.0 kmol/h B = 50.0 kmol/h TF = 92.1 oC x1D = 0.948 x1B = 0.052 TB = 108.2 oC

TD = 81.2 oC