Dp. ola geen 4A- Distillation columns can be used to separate sixtures of more than two components with the following restrictio © Tf the column has only two products, distillate and bottoms, only two product purities or recoveries can and must be specified. For each additional product stream {called side products} one more specification is required. © The McCabe-Thiele graphical design ethos is not useful to estimate the number of required stages. se notes we will introduce a short-cut rethed to estimate the reflux ratio and number of: stages, known as the Fenske- Underwood-Gilliland method, and will briefly outline how process simulators solve the rigorous distillation models. 1 Column Specifications The total number of specifications associated distillation column with two product streams is C + 6, “here C is the number of components in the feed. This includes the specifications required to define the feed to the column, which are C+ li * C- l component mole dr mass fractions * The feed cate * The feed enthalpy. Alternatively the flow of each component (C flows) and the enthalpy can be specified. The enthalpy can be specified directly or by the following two parameters:4} % The feed temperature and pressure (if the feed is a sub- cooled Liquid or @ superheated vapor) » The feed temperature of pressure and © Specification that it is a saturated Liquid or vapor © The fraction vaporized, If the column has more than one feed stream each feed stream requires C+ 1 specifications to be defined. This leaves five more specifications, to define the column design problem, These are: ® The column pressure (or pressiire profile). Ths pressure is required to establish the equilibrium relationships. Two product purity or component recovery specifications. The reflux ratie. vvy The specification that the feed be located at its optimum location (see notes on Binary Distillation Design!. The pressure profile is usually specified as either a constant pressure or as the condenser and reboiler pressures assuming a linear profile. 1.1 Key Components and Sequencing Columns Since in a multicomponent column there are more than two components and only two component purity or recovery specifications, two of the components from the mixture must be selected; these are known as the key components and are the components on which the specifications are based. The lighter of the two components is called the light key (LK) and the heavier EYis the heavy key (HK). The selection of these two components defines the way the feed mixture is cuts that is, the way the mixture is separated into the two product streams. Gbviously, most of the light key will go to the distillate and most of the heavy key will go to the bottoms. What happens to the other components? Common assumptions in solving the mass balances on the column are: © All of the components that are lighter than the light key (LEX) go to the distillate. * All of the components heavier than the heavy key (HHK) go to the bottoms. : For example, suppose that we have five components A, 8, C, D and E, listed in order of increasing normal boiling points. Figure 1 shows how the components separate if C is selected as the light key and D is the heavy key. The figure shows that components A, B and C go to the distillate and components D and E go to the bottom.fo | Distillate | A B c Feed A B | Cc (LK) D (HK) | E | } | Botoms | tO | c Figure 1. Cutting a S-component mixture into two products In the example of Fig. 1 components A and B are lighter than the light key (LLK) and component E is heavier than the heavy key (HHK). Now a small amount of the heavy key D must go to the distillate and a small amount of the light key C must go to the bottoms. How much are the small amounts depends on the product specifications, which are based on the keys, C and 0Following the same example, how many distillation columns are needed to separate the five components into five streams each sentaining one of “the components? Four is the correct an er, but how should they be sequenced? Figure 2 shows two possible ways to do the sequencing. Note that the sequencing depends on the selection of the light and heavy keys for each column. A A B A Bax 5 Cua c D . E Cay Dak) E D E L>e Figure 2. Two ways to sequence distillation columns to completely separate five components. There are 12 other ways. The following are some rules of thumb to help in sequencing distillation columns: * Remove toxic or corrosive components first to prevent the spread of toxic components of so that only the first one or two columns need to be of corrosion resistant materials. * Favor equimolal splits to keep the molar feed rates to the following columns smaller. * Remove the most plentiful component first for the same reason.* Keep difficult separations for last so that the feed rate to the column is smaller. * Remove products and recycle streams as distillates to prevent contamination with tars that may form in the bottom of the column. It is very important to always list the components in order of decreasing volatility, that is, the most volatile component at the top of the list followed by the next volatile component and so on with the heaviest component at the bottem. You must do this always. 2 Over-all Balances As with a binary distillation column, the first step is to carry out the over-all mole balances on the column to determine the flows and compositions of the product streams. With “C components, two product streams and the feed rate and composition specified, there are C independent mole balances and 2C unknowns, two product rates and 2(C - 1) product mole fractions. We have therefore C more unknowns than equations, but.. there are only the two key component specifications. This fs why we must assume that all of the lighter-than-light-key (LLK) components go to the distillate and all of the heavier- than-heavy-key (HHK) components go to the bottoms product. This gives us C - 2 zero compositions (the LLK components in the bottom and HHK components in the distillate) and we can solve the material balances. This is best demonstrated with an example.Example 1. Over-all Balances on the Distillation of a Mixture of Hydrocarbons A mixture at the rate of 200 kmole/hr is to be separated by distillation, The feed contains 10 molet propane, 45 molet n- butane, 40 mol Repentane and the balance n-hexane. It is desired to obtain a distillate stream containing no more than 4 mole’ n-pentane and a bottoms stream containing no more than 2 mole n-butane, Determine the flows and compositions of the distillate and bottoms product and the recoveries of the key components. As the specifications are given on n-butane and n= pentane, these are respectively tha light and heavy keys. So the propane is the LLK and all goes to ste distillate, so: Xa = 0 The n-hexane is the HHK, xo = 0 With four components we have four independent balances: Total balance: F = D+ B = 200 kmole/hr B= 200 -DExample 1, Over-all Balances on the Distillation of a Mixture of Hydrocarbons A mixture at the rate of 200 kmole/hr is to be separated by distillation. The feed contains 10 mole’ propa 45 molet n- butane, 40 molet n-pentai and the balance n-hexane. It is desired to obtain a distillate stream containing no more than 4 mole’ n-pentane and a bottoms stream containing no more than 2 mole’ n-butane. Determine the flows and compositions of the distillate and bottoms product and the recoveries of the key components. As the specifications are given on n-butane and n- pentane, these = are respectively the light and heavy keys. So the Propane is the LLK and all goes to. ~—sthe distillate, so: wees = 0 The n-hexane is the HHK, so: xe = 0 With four components we have four independent balances: Total balance: F = D + B = 200 kmole/hr B=200-0DComponent balances: Fxr = Dxp: + Bis: Propane: 200(0.10) = Oxocs + B(O) = 20 kmole/nr nebutane: 200(0.45) Ditsey + (200 - D10.02 = 90 kmole/hr nepentane:200(0.40) D{O.04) + (200 - Dj xan = 80 kmole/hr Sum of xy = 1 wees + Roc, + Kocy + Moco > 1-0 Hao = 160 = yey - 0.04 - 0 = 0.96 - Ape Substitute into n-butane balance: D(O.96 - xn) + (200. - B)0.02 = 90 knole/hr (0.96 - 0.02) = 90+ Dac — 2040-02) substitute propane balance: 0.940 = 90 + 20 ~ 4 = 106 kmole/be D = 106/0.94 = 112.8 smole/hr B- 200 - 112.8 = 87.2 kmole/hr nchexane balance: 200(0-05) = D(O) + (200 ~ Dixst = 10 kmole/he xeot = 10/87.2 = 0.115 acy = 1.0 - 0 - 0.02 - 0.115 = 0.865 Compositions: Distillate Bottoms: Propane: 20/112.8 = 0.177 0 n-butal 0.96-- 0.177 = 0.783 0.020 nepentanet 0.040 0.865 n-hexane? 0 0.115Recoveries: butane it aise i isat — Pepentane in mutters: i ROSA AIF the preceding @xanple the specified in the bettors prod. ane ee ee ee composition is specified in the distiliets, These are sa Fumbers and are specified this way éo the Fecowerien 9 key components are high. tit! there ba fe conver thd Aistiilece 15.184 to specify the composition of tha Lf or of the heavy key in the bottsts (4.38%) ty steute recoveries to be high. This ls why you must always specify the light key composition in the bottoms product and the heavy kay composition 10 the distillate! Having calculated the flows and carpe products, we are now ready to design determining the liquid and vapor flows ana the required number of stages. we will do this by the tt oes" method. 3 Short-Cut Distillation Design Method og is carried sat in three The short-cut distillation design © steps: ber of equilimciut stages at tote 1, Determine the minizun + x using the Penske equation.Mics, Recoveries: n-butane in distillate: 0. 783(112.8)/90 = @.98 n-pentane in bottoms: 0.865 (87.2) /80 = 0.94 In the preceding example the n-butane (LK) composition is specified in the bottoms product and the ne-pentane (HK) composition is specified in the distillate. These are small numbers and are specified this way so that the recoveries of the key components are high. Netice that there is no convenient way to specify the composition of the LK in the distillate (0.763) or of the heavy key in the bottoms (0.865) to ensure the recoveries to be high. This is why you must always specify the light key composition in the bottoms product and the heavy key composition in the distillate! Having calculated the flows and compositions of the products, we are now ready to design the column by determining the liquid and vapor flows and the required number of stages. We will do this by the short-cut design method. 3 Short-Cut Distillation Design Method The short-cut distillation design method is carried out in three steps: 1. Determine the minimum number of equilibrium stages at total reflux using the Fenske equation. 102. Determine the minimum reflux ratio using the equation. 3. Determine the number of equilibrium stages at = reflux ratio, er the reflux ratie at a specified nunber of equilibrium stages using the Gilliland correla’ These will be presented next. 3.1 Minimum Number of Stages by Fenske In the Binary Distillation Design notes we learned reflux, D= 0, B= 0: ¢ The required number of equilibrium stages is mi: * The operating lines become Yau = xa- Assuming that the relative volatility a, of component i to component j is constant, a formula can be derived to calculate the minimum number of stages at total reflux. This form known as the Fenske equation. The following is the derivation. Histjn Yin za = or =a, ay Equilibrium relation: ay = 28 or Je Operating line: Combine A and B: ay Spee D: At the top of the column: Yi =X E: From C: 2 a BE = gy Bm oy Ba = fmin Timi x ay HD Xp iM pun At the bottom of the column: wi. =%e 1mom ‘min ann BR oe afin Gt Substitute Fin Ey ie va "0 im Hi Take logarithm: Neil (eh) = in( for Nua to obtain the Fenske equation: x in( 228 N, ni qa min In(aiy) The derivation is presented here to show that there is no “- 1” The “- 1* on the equation as most textbooks and handbooks show. is supposedly added to account for a partial reboiler which is not possible because at total reflux both the condenser and the reboiler must be total. The Fenske equation applies to any two components {in the multicomponent mixture, but to calculate the minimum number of stages we must use the two key components. Why? Because the '§ are given on the key components. product purity specification the equation once the minimum number of stages is determined, used to estimate the compositions of the non-key in the bottems and distillate products, the ones can be components assumed to be zero in the mass balances, at total reflux. Equation (1) shows that the minimum number of stages depends t compositions and the relative wolatility of only on the produc and independent of the feed composition the two key components, and enthalpy, as is the case with binary columns. 12the Wnen the relative volatility is only approximately constant geometric average of the values at the top and bottom of the column can be used: Syavg = of Fyrap%j.v0r Note that the relarive volatilities at the top and bottom are for the same pair of components, not for a different pair. Example 2, Minimum Number of Stages for a Mixture of Hydrocarbons The velative volatilities for the components of Example 1 are given at the top and bottom of the column in the table below, along with the results for the specifications of Example 1. Determine the minimum number of equilibrium stages using the Fenske equation. Determine also the compositions sf the non-key components at total reflux. Component XP Xal x, @ top) a, bat propane o10 0177 0.000 775) ee $$} — el aol aco a: butane 0.45 0.783, 0.020 460 1.000 calculated relative to Note that the relative volatilities are the heavy key, n-pentane. Solution: Compute the average relative volatilities: 13ae Propane: ay = {(6.775)(1L.! = 6.671 nebutane: yy = ¥(2460)G.172) = 2, nohexane: ay = ¥(0422)(0.330) = 0. Minimum number of stages: Ny To obtain the compositions o reflux we apply the Fenske equ: Propane (LLE): Nmin = inte) A 793 373 eB) _ (ass 13)(0.65) 1910-090) — 6.6 in = tntuzs) «sn. 793) f the non-key compon ation to those compon! tn(S0se2) n(iguemusae) (0.41. 7)(0.865) ) 0020 xj Rae 671) ante at ents Nga = OE OBEE) g 6717-86 = 3.2 x 107° 0.020 n-hexane (HAR): Nnin = wes) (zat) = a = 6.6 In(ecaa) ‘in(0.373) Xpg = CEMOU 9.37366 = 2.0 x 10 865 Both of these mole fractions were assumed to be ze! palances of Example 1. They are small enough so balance results are not affected. It is important to realize involve approximations such accurate to at most three si ro in the mass that the mass that, because design calculations as these, the results are only gnificant digits. This is why you must not report your results to more than three digits.3.2 Minimum Reflux Ratio—The Underwood Equation The minimum reflux ratio, as with binary distillar depends on the enthalpy and composition of the relative volatilities {equilibrium relations) and the ais composition. The Underwood equation provides a liate quick way estimate the minimum reflux ratio; it requires two steps. 1. Determine the value of @ that satisfies the 1-q= 35 3) 2. Use the value of @ to calculate the minimum ref Rom 1 (4) Where a, = relative volatility of component i q = enthalpy condition parameter of the feed xpy = mole fraction of component i in the feed Xi mole fraction of component i in the distillate ‘Ro = minimum reflux ratio The enthalpy parameter g is presented in detail in the Binary stillation Design notes. 15 epEquation (3) must be solved iteratively, The value of 2 can be determined with an appropriate calculator, a workbook or a Program such as MathCad or MATLAB. When these are not available the Secant method can be used (see the Flash Distillation Rotes). However, in this case the iterative solution presents a Particular problem, regardless of the solution method used. For a system with © components Eq. (3) is a polynomial of degree C with C different roots, but only one of thase roots is valid for use in Eq. (4). How te ensure to obtain the correct value of 2? By starting the iteration with a value of @ which is between the relative volatilities of the two key components. When this is done the resulting value will be between the two relative volatilities and this is the valid reot to use in £q. (4). Example 3. Minimum Reflux Ratio for a Mixture of Hydrocarbons For the conditions of Example 1 determine the minimum reflux ratio. The feed enters the column 40% vaporized, by mole, at the conditions of the feed. Solution. The average relative volatilities were determined in Example 2. As q is the molar fraction of the feed that is Liquid, q=1- 0.40 = 0.60 Equation 3 te solve for @: 8.671(0.10) | 2.793(0.45) , 1.0(0.40) _ 0.373(0.05) 1°060= Gen-@ * 3793-0 * 10-0 * aso Secant iteration, The value of @ is between 1.0 and 2.793, 16First guess: % = 2.0 0.40 + MEZHOAM) 4, 27OMOAS) | LoC04d) | ox7s(a95) _ 618 040+ Tiinae * aoe-ae * tee + oaricay ~ OF LS ey = 0.40 + SAGER) 4, Brenna) Uae Secant: = 15 — (0.124) 28 = 1.560 es) = 0.011 = 1.555 e.= 0.0006 The minimum reflux ratio from Eq. (4) with @ = 1.555: Rome azHOHiT) 4 27900703) , Lokanany , _wsTaCE)_ Rava-1555 * T7H=1555 * 10-1555" OITS-1SSS 0.837 In this example the Secant method converges within four iterations. 3.3 Number of Equilibrium Stages—Gilliland Correlation With the minimum number of stages at total reflux and the minimum reflux ratio, the Gilliland correlation allows the calculation of the required number of equilibrium stages at a specified reflux ratio, or the reflux ratio required for a specified number of equilibrium stages. The original correlation was graphical, shown in Fig. 3. 17i (a l/iW #2) 100 0.010 I - - 08 oxo 100 | Mealy #2) Figure 3. Gilliland Correlation The correlation has been converted to the following formula for ease of programming: Imin ( 14544XK_\X-1 SeMoin — 4 — glirerizan, wal # (5) . — Ro-Rpm Ww = Sorkpe here Xx Roti N= number of equilibrium stages 18 iNoun = minimum number of stages at total reflux Ro = reflux ratio Roe = minimum reflux ratio 3.4 Optimum Location of the Feed Tray The optimum location of the feed stage may be estimated by the Kirkbride equation: slemaee™ N-Ne xetK) \xpu) D. Where N, = number of stages above the feed tray, N = number of equilibrium stages B = bottoms product molar rate D = distillate molar rate Xe.un Neue = mole fractions of the keys in the feed Mace * mole Eraction of the light key in the bottoms Moan = mole fraction of the heavy key in the distillate Equation 6 does net provide good estimates of the optimum feed tray location when the relative volatility of the key components 19differs significantly from the top to the battom of the column, or when they form an azeotrope. Example 4. Number of Equilibrium Stages for Hydrocarbon mixture Determine the number of actual trays required for the column of Example 1 at 2 reflux ratio of 1.2 times the minimum ratio and for an over-all column efficiency of 50%, Assume the column has a total condenser and a partial reboller. Determine also the optimum location of the feed tray- Solution. The minimum number of stages was calculated in Example 2 and the minimum reflux ratio in Example 3. Nein = 6.6 Roy = 0.837 Reflux ratio: Rp = 1-2(0,837) = 1-004 R om — 1004-0857 — 0.0835 Rp+1 1.00441 from Fig. 3: “kms = 0.57 N= Assuming the partial reboiler is an equilibrium stage, the number of actual trays required for 50% over-all efficiency is: N, = (16.7 - 1)/0.50 = 31.4 of 32 actual trays. The Gilliland correlation can also be used to estimate the reflux ratio required for a specified number of equilibrium stages. 20Optimum feed tray location, “mele ae) (ee see ag Ne = N/(1 * 1.19) = 16.7/2.29 = 7.3 in Hs So, 7.3 stages above the feed and 9.4 below. In real trays efficiency); Above the feed tray: 7.3/0.50 = 14.6 of 15 trays Below the feed tray: (9.4 - 1)/0.50 = 16.8 of 17 trays We subtracted one stage from the bottom to account for the Partial reboiler 4 Rigorous Distillation Models The method presented in the preceding section, sometimes known as the Fenske-Underwood-Gilliland method, is used by process simulators to estimate the number of equilibrium stages at a specified reflux ratio in models also known as Short-cut Distillation Models. Simulators also provide rigorous distillation models that solve the mass and enthalpy balances and equilibrium relationships on each tray. We will briefly present here the equations involved te give an idea of the complexity of the solution procedure. It is important to understand that the rigorous method is a rating method rather than a design method. For the rigorous method, instead of specifying the reflux ratio we must specify the number of equilibrium stages and the location of the feed 21ye tray. In other words, the method is designed to rate a prop: umn. TI column. The five specifications required to solve the a rigorous model are: Sprouse 1. The number of equilibrium stages 2. The lecation of the feed stage 3. The pressure profile 4, Either the reflux or beil-up ratio, the distillate purity or the light key component recovery 5. Either the distillate or bottoms product flow, the bottoms product purity or the heavy key component recovery. Minimum reflux and minimum number of trays have no meaning the rigorous method. bea Ve Kiel Vik | Hoes Hux | Figure 4. Schematic of a Tray in a Distillation Column 22 inThe rigorous model calculations are performed o1 Consider the tray sketched in pig. 4 n each tray. results in: = A total molar balance Loa tH = Ly +h, 1 legn., 2 unks. [Li Vi] Where L and V are the molar rates of the liquid and vapor streams (kgmole/hr), respectively, and the subscript k refers to the tray number. Next we write component molar balances for each of the components minus oni Laskin t hinYian = Lamia + Vs i) Ceqns., 26 + 2 unks. [xi,e- viel Where x and y are the component mole fractions in the liquid and vapor streams, respectively. The total unknowns include all c components, so we can write: (9) © +2 eqns., 2¢ + 2 unks. 23wring the tray “San equilib gre related by: rium stage; Ya = KT Pox, (10) 2¢ + 2 eqns., 20+ 3 unks. [Te] Where 7, is the temperature on the tray (K) and P is the pressure (kPa). In counting unknowns we have considered that the K-factor is calculated by the simulator from the specified thermodynamic model, and the pressure is either specified constant or as linearly varying between the specified condenser and reboiler temperatures. Finally an enthalpy balance completes the model of the tray: LHe Vettes Ha (ua) 2¢ + Deqns., 2¢ + 3 unks. Solved! knere H: and He are the specific enthalpies of the liquid and wapor streams (J/kgmolel, respectively. These enthalpies are calculates by the simulator as functions of the temperature, pressure and composition using the specified thermodynamic package. Notice that the solution of the tray equations requires the iter ve solution of the flash equations, similar ta those solves in the notes on flash Distillation. Furthermore, because 24ne inlet streams to the tray are out above and the tray below let streams from the tray equations the complete solution of the column requir — ®S very sophisticated iteration algor mlators possess these algorithms and are very us for solving these equations 7 Summary These notes have dealt distillation columns. with the design of multicompon The need for the definition of key components was introduced and the ‘short cut design method was presented in detail. This included the estimation of the minimum reflux ratio at total reflux by the estimation of the minimum Fenske equation, the reflux ratio by the Underwood equation, and the estimation of the number of equilibrium stages at a specified reflux ratio by the Gilliland correlation. A brief glance at the equations involved in the rigorous distillation models available in process simulators is included. also Review Questions 1. How many specifications are needed to define a column with C components in its feed stream and two product streams? How many of these are associated with the feed stream? 252.List the fiv, © column specifications not associated with the feed stream required t och sates, ‘@ design the column by the short- Why is it A it necessary to select key components to design a multicomponent distillation column? 4.In ord: tder to solve the mass balances on a distillation col ‘Olumn, what assumptions must be made on the lighter-than— light-key {LLK) components and on the heavier-than-heevy- key (HHK) components? S.In the short-cut design method, for what is the Fenske equation used? For what is the Underwood equation used? For what is the Gilliland correlation used? 6. List the five column specifications that must be made to solve the rigorous distillation model. Why is this model called a rating model? Problems 1. Design of a ‘Multicomponent Distillation Colum. The feed to a conventional distillation column is 30% vaporized at the conditions ef the feed plate. Given below are the feed composition and the relative volatilities of the components. It is desired to have no tore than 2 moles of component ¢ in the distillate, and no more than 1 molet of component B in the bottoms product. Draw a schematic of the column showing a11 the problem data on the diagram and, using’ # reasonable reflux ratio,determine the number of actual stages required for an over-all efficiency of 608, @ total condenser and 4 partial reboiler. Report also the location of the feed stage, the flows and compositions of 26