McCabe-Thiele Diagrams for Binary Distillation

Tore Haug-Warberg Dept. of Chemical Engineering August 31st, 2005

The McCabeThiele diagram for solving bi < 0 nary distillation problems represents a pillarQ stone in every chemical engineering class. The V1 reason for this is not its practical importance which in the era of modern computer technolD L0 V2 ogy has become very modestbut its educaL1 tional value based on a very illustrative graphical picture of a complicated physical process. The McCabeThiele diagram in its most simple Vn form represents a graphical solution to the clasLn1 F sical separation problem of N + 1 ideal equilibVn+1 rium stages connected with countercurrent vaLn por and liquid ows as illustrated in Fig.1. The streams are shown as separated ows except at certain distinct locations (trays) where complete mixing and thermodynamic equilibrium is assumed. This is a useful abstraction but it must VN be understood that the situation in a real plant LN 1 V deviates in two respects: First of all there is N +1 > 0 no one-to-one correspondance between a physiQ cal tray (or plate) and a thermodynamic equilibB LN rium stage1. Secondly, many columns have their plates replaced by a structured packing which F 1: Simplied picture of a counter-current gives intimate contact between the vapor and liquid phases at all locations in the column. The distillation column. simplied picture is still valid, however, if each tray is considered to be one theoretical equilibrium stage regardless of the physical realization of it.
ical equilibrium stage reached for each physical plate. This eciency may reach values as high as 0.80-0.90 but is in many cases much lower.
1
1It is common practice to talk about tray eciencies which is the fraction of a theoret-

1. The energy balance To calculate the solution of the complicated separation problem in Fig.1 we need 3 sets of mass balances, one for the rectifying, one for the feed and one for the stripping section of the column. In addition there is an overall energy conservation assumption known as constant molar overow. The origin and importance of this assumption is clearied below. Let i n represent any of the internal stages in the the column except the feed tray. The energy balance for an adiabatic stage is vap + H liq = H vap + H liq H
i+1 i1 i i

Since enthalpy is an Euler homogenous function of rst order (extensive state variable) the energy balance can also be written
liq liq vap hvap i+1 Vi+1 + hi1 Li1 = hi Vi + hi Li

where h, and V and L denote the molar enthalpy and molar ows of the vapor and liquid streams respectively2. Combined with the mass balance Vi+1 + Li1 = Vi + Li for stage i the last equation becomes
vap vap liq vap hvap hliq i1 hi+1 Li1 = hi i+1 Vi + hi hi+1 Li vap Adding and subtracting hvap i1 Li1 and hi Li brings the equation on the following form: vap vap vap vap hliq hvap i1 hi1 Li1 + hi1 hi+1 Li1 = hi i+1 Vi vap + hliq hvap Li + hvap i hi i i+1 Li

The enthalpy function has no absolute zero, and it is only the enthalpy differences that matter in the energy balance. But, the importance of the temperature dependency of enthalpy is usually much smaller than the heat of vap = hvap vaporization3. The assumption that hvap i1 = hi i+1 is therefore quite legitimate and the energy balance can be simplied to:
vap liq vap hliq Li i1 hi1 Li1 = hi hi

Finally, assuming that the heat of vaporization is constant for the two-phase vap liq vap mixture makes hliq and consequently Li1 = Li . The last i1 hi1 = hi hi argument can easily be extended to yield L0 = L1 = L2 = . . . = Ln1 V1 = V2 = V3 = . . . = Vn
This is according to common practice in this eld. 3 Typical changes per stage are 100500J/mol for sensible heats and 25000 35000J/mol for heats of vaporization
2The dot-notation has been dropped for the rate symbols.

2. OPERATING LINES

which is known as the constant molar overow assumption4. The same assumption is valid for the bottom section as well, but not across the feed tray, where a discontinuity occurs because of the feed stream entering from outside the control volume. 2. Operating lines This section (and the rest of the paper) is restricted to the distillation of binary mixtures. In the top section of the column an enrichment of the lighter of the two components takes place and in the bottom section the heavier component is enriched5. The next task is to express the mass balances of the column and have them integrated with the energy balance from the previous section into something called an operating line. The composition of the lighter component is designated by y and x for the vapor and liquid phases respectively. By convention these symbols stand for the lighter component. 2.1. The rectifying section. We shall write the mass balances as one total balance and one component balance. For the top section these equations are Vi = D + Li1 yi Vi = x0 D + xi1 Li1 which means the heavier component is never considered explicitly6. The equations are easily solved for the vapor composition: D Li1 yi = x0 + xi1 Vi Vi D Li1 = x0 + xi1 D + Li1 D + Li1 The constant molar overow assumption yields L0 = L1 = . . . = Ln1 L. Furthermore, it is customary to dene the external reux ratio of the column as R L/D [0, : 1 R + xi1 , i [1, n] yi = x0 1+R 1+R This is the so-called operating line representing the combined mass and energy balances in the rectifying section of the column. The equation is valid for all of the internal stages i = 1, 2, . . . , n and in particular for i = 1 which yields y1 = x0 = xD . A bit surprising maybe,
for any number of the chemical components in the mixture. means that the temperature prole of the column decreases from the bottom towards the top. 6 For binary mixtures it can easily be eliminated from the given equations.
5This 4Valid

y1 y2 y3 y4 slope =
R 1+R

x0 1+R

x3

x2

x1

xD

F 2: Rectifying section of a benzene toluene column (R = 2).

but from Fig.1 we see that the condenser has no separating eectit only condenses the vapor into a liquid phase7, 8. For i = 2 the operating line connects the vapor composition from stage 2 with the liquid composition from stage 1, but here the calculation stops. We cannot move on before the liquid composition from stage 1 is known! The answer to this problem lies in the assumption of thermodynamic equilibrium from which we can derive x1 = xeq (y1 ), x2 = xeq (y2 ) and so on. With this relationship in mind we got all three ingredients of the McCabeThiele graphical method: (1) Constant molar overow assumption (2) Operating line (3) Equilibrium curve The diagram is conventionally drawn with the heavier component to the left and the lighter component to the right. That means the equilibrium curve is located above the main 45 diagonal in the diagram. Fig.2 shows a typical example for an almost ideal mixture of benzene (light) and toluene (heavy).
This is referred to as a total condenser. In some cases the column is equipped with a partial condenser which makes xD x0 . The condenser must then be treated like an additional equilibrium stage, see the discussion of the stripping section below. 8 Prove that the heat duty of the condenser is in the range R Q/(Dvap h) R + 1 for partial versus total operation.
7

2. OPERATING LINES

The staircase (red) shows the interaction between the equilibrium curve (vertical step) and the operating line (horisontal step). Note that the operating line (black) crosses the equilibrium line (blue) at one point in the diagram. The column cannot operate to the left of this cross-over point because that would be in the one-phase region of the phase diagram. Hence, the number of theoretical stages approaches innity as the column strives towards a singular condition9. 2.2. The stripping section. The operating line for the bottom section follows the same receipe as for the top section. The two mass balances are principally the same, Vi + B = Li1 yi Vi + xB B = xi1 Li1 and again we solve for the vapor composition on the left hand side: B Li1 yi = xB + xi1 Vi Vi B Vi + B = xB + xi1 Vi Vi The constant molar overow assumption yields Vn+1 = Vn+2 = . . . = VN +1 V . Furthermore, we may characterize the heat duty10 of the reboiler as S V/(V + B) 0, 1] and write 1S 1 + xi1 , i [n + 1, N + 1] yi = xB S S where S is the fraction of the downcoming liquid which is being evaporated. Alternatively we can dene S as the heat duty of the reboiler divided by the heat which is needed to evaporate all the downcoming liquid. The operating line is valid for all the internal stages i = n + 1, n + 2, . . . , N and also for the reboiler i = N + 1. The stripping section has therefore one external equilibrium stage in addition to the internal stages of the column (the situation is similar to that of a partial condenser in the rectifying section). The cross-over between the operating line and the main diagonal is given by yN +2 = xN +1 = xB . The index i = N + 2 is strictly speaking outside the control volume but it does still serve as an easy-toremember construction point of the diagram, see Fig.3. At this point it is possible to count the number of theoretical stages from the specications of the top and bottom products together with the condenser and reboiler duties. It is more common, however, to specify
This is barely visible in the current diagram but an innitly large magnifying glass would reveal an innite number of successively smaller steps. 10 Prove that the heat duty of the reboiler is Q = S (1 S )1 Bvap h.
9

y5 y6 y7 y8

slope =

1 S

xB x7 x6 x5 x4 F 3: Stripping section of a benzene toluene column (S = 2/3). Tray numbers are rounded to the nearest integer.

the feed stream rather than the reboiler duty. To manage this we need the operating line for the feed tray which is discussed in the next section. 2.3. The feed tray. The operating line for the feed tray (q-line) requires four mass balances and two energy balances (including the constant molar overow assumption). The extra mass and energy balances are needed to characterize the physical state of the feed stream. To begin with we have the component balances yn Vn = xD D + xn1 Ln1 xn Ln = yn+1 Vn+1 + xB B for the rectifying and stripping sections respectively. At the feed tray there will in general be sudden jumps in all the stream variables, but for an optimal placement of the feed tray the intensive state variables (temperature and composition) are continuous. Hence, we shall require that yn = yn+1 y and that xn = xn1 x in the equations above. Summation plus reorganization yields: y (Vn+1 Vn ) = x (Ln Ln1 ) ( xD D + xB B)
xF F

2. OPERATING LINES

Note that the total mass balance xF F = xD D + xB B is substituted on the right side. The q-line is rst written on the preliminary form: Ln Ln1 F y=x xF Vn+1 Vn Vn+1 Vn The feed stream is subsequently replaced by the mass balance for the feed tray F = (Ln Ln1 ) (Vn+1 Vn ). A little manipulation yields y=x Ln Ln1 Ln Ln1 xF 1 Vn+1 Vn Vn+1 Vn L L x xF 1 V V

where V and L measure the dierences in the vapor and liquid ows across the feed tray (top-down). This could serve as our nal result, but it is customary to characterize the feed quality by L q L V where q = 1 V = 0 (saturated liquid feed) and q = 0 L = 0 (saturated vapor feed). Hence, the feed quality can be interpreted as the fraction of saturated liquid in the feed. The denition of the feed quality is easily inverted to give L/V = q/(q 1) which puts the q-line on the nal form: q 1 y=x xF q1 q1 This is the last of the three operating lines used in a McCabe-Thiele diagram, but it should be mentioned that the optimal feed tray location causes a slight redundancy in the calculation of the diagram: E.g. in most cases the composition specications xF and xD are given in addition to the reux ratio R and the feed quality q. This species the upper part of the column and the conditions of the stripper will be xed once xB or S is known (the other parameter is xed by the common cross-over point of the operating lines). Anyway, the calculation is complete in the sense that it reveales all the signicant gures of a binary distillation column (see also Fig.4): (1) Compositions of the top and bottom products (2) Composition and quality of the feed stream (3) Reux ratio (and condenser duty) (4) Reboiler duty 2.4. Extreme specications. The condition of minimum reux is reached then i) the cross-over between the operating lines is located at the phase boundary as illustrated in Fig.2 or ii) the upper operating line is tangent to the equilibrium curve while the cross-over point is inside the phase diagram

1 0.8 Ntop = 4.03 0.6 0.4 Nbot = 4.42 0.2 0 0 slope =

q q1

0.2

0.4

0.6

0.8

F 4: Complete McCabe-Thiele diagram for a benzenetoluene column with parameters xD = 0.9, xF = 0.55, xB = 0.1, R = 2 and q = 0.2 (consistent with S = 2/3).

(applies to mixtures with concave equilibrium curves). In both cases the number of theoretical stages approaches innity. The other extreme is then the reux is innite and the operating lines for the column overlap with the main diagonal. This yields the smallest number of theoretical stages for a given separation. There are also cases when the stripping section or the rectifying section is missing or ineective. These cases include batch distillation (no stripper) and degassing units (no rectifyer). Mathematically, there is also the possibility that a section becomes ineective because of an extreme feed condition. This is readily veried from the q-line which has slope 1 for q = . The rst case (q = ) corresponds to a liquid feed so cold that it condenses all the vapor coming up from the reboiler (ineective rectifying section). The second case (q = ) corresponds to a superheated vapor so hot that it evaporates all the liquid coming down from the condenser (ineective stripping section).