Chapter 6:

Approximate Shortcut Methods for

Multi-component Distillation

Total Reflux: Fenske Equation

For a multi-component distillation with

total reflux as shown in the Figure below

1
 The vapour-liquid equilibrium equation
for any two components, A & B, at the
re-boiler can be written as follows
⎛ yA ⎞ ⎛ xA ⎞
⎜ y ⎟ = αR ⎜ x ⎟ (6.1)
⎝ B ⎠R ⎝ B ⎠R

Note that Eq. 6.1 is, in fact, the

definition of relative volatility of species A
and B
yA x A
α AB = (6.2)
yB x B

Performing the material balances for

species A around the re-boiler gives
Lx A ,N = Vy A ,R + Bx A ,R
or
Vy A ,R = Lx A ,N − Bx A ,R (6.3)

2
Doing the same for species B yields
VyB ,R = Lx B ,N − Bx B ,R (6.4)

Since this is a total reflux distillation,

B =0
V =L

Eq. 6.3 can then be re-written as follows

Vy A ,R = Lx A ,N
y A ,R L
= =1
x A ,N V
y A ,R = x A ,N (6.5)

Doing the same for Eq. 6.4 results in

yB ,R = x B ,N (6.6)

3
 Eqs 6.5 & 6.6 confirm that the operating
line for the total reflux distillation is the
y = x line

Combining Eqs 6.5 & 6.6 with Eq. 6.1

gives
⎛ yA ⎞ ⎛ xA ⎞
⎜ y ⎟ = αR ⎜ x ⎟
⎝ B ⎠R ⎝ B ⎠R
⎛ xA ⎞ ⎛ xA ⎞
⎜ x ⎟ = αR ⎜ x ⎟ (6.7)
⎝ B ⎠N ⎝ B ⎠R

For stage N (in stead of stage R or at the

re-boiler), the vapour-liquid equilibrium
equation can be written as follows
⎛ yA ⎞ ⎛ xA ⎞
⎜ y ⎟ = αN ⎜x ⎟ (6.8)
⎝ B ⎠N ⎝ B ⎠N

4
 By performing material balances for
species A and B around stage N, we obtain
y A ,N = x A ,N −1 (6.9)
yB ,N = x B ,N −1 (6.10)

Accordingly, Eq. 6.8 can be re-written,

by combining with Eqs. 6.9 & 6.10, as
follows
⎛ xA ⎞ ⎛ xA ⎞
⎜x ⎟ = αN ⎜x ⎟ (6.11)
⎝ B ⎠N −1 ⎝ B ⎠N

Combining Eq. 6.11 with Eq. 6.7 yields

⎛ xA ⎞ ⎛ xA ⎞
⎜x ⎟ = αN αR ⎜ ⎟ (6.12)
⎝ B ⎠N −1 x
⎝ B ⎠R

5
 For stage N − 1, we obtain
⎛ xA ⎞ ⎛ xA ⎞
⎜x ⎟ = α N −1α N α R ⎜ ⎟ (6.13)
⎝ B ⎠N −2 x
⎝ B ⎠R

When we repeat the similar derivations

until we reach the top of the distillation (i.e.
stage 1), we obtain
⎛ xA ⎞ ⎛ xA ⎞
⎜ x ⎟ = α1α 2α 3 ...α N −1α N α R ⎜ x ⎟
⎝ B ⎠dist ⎝ B ⎠R
(6.14)

If we define α AB as the geometric-average

relative volatility,
1
α AB = (α1α 2α 3 ...α N −1α N α R )N min

where N min = # of equilibrium stage for the

total reflux distillation,

6
Eq. 6.14 can be re-written as
⎛ xA ⎞ N min ⎛ x A ⎞
⎜ x ⎟ = α AB ⎜ x ⎟ (6.15)
⎝ B ⎠dist ⎝ B ⎠R

Solving for N min results in

⎡⎛ x A ⎞ ⎤
⎢⎜ x ⎟ ⎥
ln ⎢ ⎝ B ⎠dist ⎥
⎢ ⎛ xA ⎞ ⎥
⎢ ⎜x ⎟ ⎥
N min = ⎣ ⎝ B ⎠R ⎦ (6.16a)
ln α AB

Eq. 6.16a can also be written for multi-

component distillation as follows
⎡ ⎛ Dx A ⎞ ⎤
⎢ ⎜ Dx ⎟ ⎥
ln ⎢ ⎝ B ⎠dist ⎥
⎢ ⎛ Dx A ⎞ ⎥
⎢ ⎜ Dx ⎟ ⎥
N min = ⎣ ⎝ B ⎠R ⎦ (6.16b)
ln α AB

7
 In the previous Chapter (hand-outs), we
have learned that
Dx A , dist = ( frac. recov. in dist.) Fz A
(6.17)

Let
frac. recov. of A in dist. = ( FR A )dist

Eq. 6.17 can be re-written as

Dx A , dist = ( FR A )dist Fz A (6.18)

Also from the previous Chapter (hand-

outs),
Bx A ,R = ⎣⎡1 − ( FR A )dist ⎦⎤ Fz A (6.19)

Note that Bx A ,R ≡ Bx A , bot

8
 For species B, we can also write the
similar equations as Eqs. 6.18 & 6.19 (try
doing it yourself)

Combining Eqs. 6.18 & 6.19 and the

corresponding equations for species B with
Eq. 6.16, and re-arranging the resulting
equation gives
⎧ ( FR A )dist ( FRB )bot ⎫
ln ⎨ ⎬
⎩ ⎡⎣1 − ( FR A )dist ⎤⎦ ⎡⎣1 − ( FRB )bot ⎤⎦ ⎭
N min =
ln α AB
(6.20)

Note that ( FRB )bot = fractional recovery of

species B in the bottom product

9
 When there are 2 components (i.e. a
binary mixture), Eq. 6.20 or Eq. 6.16 is
reduced and re-arranged to
⎧ x A (1 − x A )dist ⎫
ln ⎨ ⎬
⎩ x A (1 − x A )bot ⎭
N min = (6.21)
ln α AB

Note that A = more volatile component

(MVC)

Eq. 6.20 can also be written for C and B

(C = a non-key component but B = a key
component) as follows
⎧ ( FRC )dist ( FRB )bot ⎫
ln ⎨ ⎬
⎩ ⎡1 − ( FRC )dist ⎦⎤ ⎡⎣1 − ( FRB )bot ⎤⎦ ⎭
⎣
N min =
ln αCB
(6.22)

10
Solving Eq. 6.22 for ( FRC )dist results in

αCB
N min

( FRC )dist =
⎡ ( FRB )bot ⎤
⎢ ⎥ + α N min

⎣1 − ( FR )
B bot ⎦
CB

(6.23)

11
Example An atmospheric distillation
column with a total condenser and a partial
re-boiler is used to separate a mixture of 40
mol% benzene, 30% toluene, and 30%
cumene. The feed is input as a saturated
vapour. It is required that 95% of toluene
be in the distillate and that 95% of cumene
be in the bottom. If the CMO is assumed
and the reflux is a saturated liquid,
determine a) number of equilibrium stage
for total reflux distillation and b) fractional
recovery of benzene in the distillate
Given the constant volatilities with
respect to toluene as α benz-tol = 2.25 and
α cume-tol = 0.21

12
 From the given information, toluene =
LK, cumene = HK, and benzene = LNK

Note that the order of volatility is as

follows
α benzene > α toluene > α cumene

Let assign that toluene = A, cumene = B,

and benzene = C

Thus,
• α benz-tol = αCA = 2.25
• α cume-tol = α BA = 0.21

1
Note that α AB = (try proving this
α BA
yourself)

13
 Also from the given data, we obtain the
following:
• z A = z toluene = 0.30
• z B = z cumene = 0.30
• zC = z benzene = 0.40
• ( FR A )dist = 0.95

• ( FRB )bot = 0.95

The number of minimum equilibrium

stages can be computed, using Eq. 6.20, as
follows
⎧ ( FR A )dist ( FRB )bot ⎫
ln ⎨ ⎬
⎡1 − ( FR ) ⎤ ⎡1 − ( FR ) ⎤
= ⎩⎣
A dist ⎦ ⎣ B bot ⎦ ⎭
N min
ln α AB
⎧ ( FR A )dist ( FRB )bot ⎫
ln ⎨ ⎬
⎩ ⎡
⎣1 − ( A )dist ⎦ ⎣ ( B )bot ⎦ ⎭
FR ⎤ ⎡1 − FR ⎤
=
⎛ 1 ⎞
ln ⎜ ⎟
α
⎝ BA ⎠
14
 ⎧
ln ⎨
( 0.95 )( 0.95 ) ⎫
⎬
N min = ⎩ [1 − ( 0.95 )] [1 − ( 0.95 )] ⎭
⎛ 1 ⎞
ln ⎜ ⎟
⎝ 0.21 ⎠
= 3.8

We need the value of αCB to calculate

fractional recovery of species C (benzene) in
the distillate, but we have the value of αCA ,
how do we obtain the value of αCB

Note that αCB is, in fact,

KC
αCB = (6.24)
KB

Dividing both numerator and denomi-

nator of the RHS of Eq. 6.24 with K A yields

15
 KC K A
αCB = (6.25)
KB KA

Since K C K A is, in fact, αCA , and K B K A

is α BA , the value of αCB can be calculated as
follows
αCA 2.25
αCB = = = 10.7
α BA 0.21

Thus, the fractional recovery of benzene

(species C) in the distillate can be
computed, using Eq. 6.23, as follows
αCB
Nmin

( FRC )dist =
⎡ ( FRB )bot ⎤
⎢ ⎥ + α N min

⎣ ( B )bot ⎦
CB
1 − FR
(10.7 )
3.8

=
⎡ ( 0.95 ) ⎤
⎢1 − 0.95 ⎥ ( )
3.8
+ 10.7
⎣ ( )⎦
( FRC )dist = 0.998
16
Minimum Reflux: Underwood Equations

For a binary (2-component) mixture, the

pinch point usually (but NOT always)
occurs that the feed stage/plate, as shown
below

Note that the exception of the pinch

point that does not occur at the feed stage
can be illustrated in the Figure on the next
Page
17
 The assumption that the pinch point
occurs at the feed stage can be used in most
cases (except for some special cases, such
as an azeotropic distillation) and can also be
extended to multi-component systems

18
 A.J.V. Underwood (1948) developed an
analysis procedure to find the minimum
reflux ratio, which comprises a number of
equations

The development of Underwood

equations is quite complex, and it is not
necessary, especially for practicing
engineers, to understand all the details of
the development

Thus, we shall follow an approximate

derivation of R.E. Thompson (1980), which
is good enough for engineering calculations

19
 Consider the enriching/rectifying section
of a distillation column as follows

Performing a material balance for

species i at the minimum reflux ratio gives
Vminyi , j +1 = L min x i , j + Dx i , dist (6.26)

20
 At the pinch point (for the case that the
pinch point is at the feed stage), since it is
on the equilibrium curve, the compositions
are constant; thus,
x i , j −1 = x i , j = x i , j +1 (6.27a)

and
yi , j −1 = yi , j = yi , j +1 (6.27b)

The equilibrium equation of species i at

stage j + 1 can be written as follows
yi , j +1 = K i x i , j +1 (6.28)

Combining Eq. 6.26 with Eqs. 6.27 (a+b)

and 6.28 results in
yi , j
Vminyi , j +1 = Lmin + Dx i , dist (6.29)
Ki

21
 Let define the relative volatility, αi , as
follows
Ki
αi = (6.30)
K ref

Combining Eq. 6.30 with Eq. 6.29 and

re-arranging the resulting equation yields
yi , j
Vminyi , j +1 = Lmin + Dx i , dist
αi K ref
yi , j
Vminyi , j +1 − L min = Dx i , dist
αi K ref
⎛ Lmin ⎞
Vminyi , j +1 ⎜1 − ⎟ = Dx i , dist
⎝ Vminαi K ref ⎠

Dx i , dist
Vminyi , j +1 = (6.31)
⎛ Lmin ⎞
⎜1 − V α K ⎟
⎝ min i ref ⎠

22
 Multiplying both numerator and denomi-
nator of the RHS of Eq. 6.31 gives
αi Dx i , dist
Vminyi , j +1 = (6.32)
⎛ Lmin ⎞
⎜ αi − V K ⎟
⎝ min ref ⎠

Taking a summation of Eq. 6.32 for all

species results in
αi Dx i , dist
∑V min yi , j +1 = Vmin = ∑
⎛ Lmin ⎞
⎜ αi − V K ⎟
⎝ min ref ⎠

(6.33)

Doing the same for the stripping section

yields
αi Bx i , bot
−Vmin = ∑ (6.34)
⎛ Lmin ⎞
⎜ αi − V K ⎟
⎝ min ref ⎠

23
 Note that, since the conditions in the
enriching/rectifying section are different
from those in the stripping section,
normally,
αi ≠ αi
and
K ref ≠ K ref

Underwood defined the following

Lmin
φ= (6.35a)
Vmin K ref
Lmin
φ = (6.35b)
Vmin K ref

24
 Thus, Eqs. 6.33 & 6.34 become
αi Dx i , dist
Vmin =∑ (6.36)
(α i − φ )
and
αi Bx i , bot
−Vmin =∑ (6.37)
(α i − φ )

If there are C species (components),

there will be C values (or roots) for each φ
or φ

(6.36) + (6.37) gives

⎡αi Dx i , dist αi Bx i , bot ⎤
Vmin − Vmin = ∑⎢ + ⎥
⎣ ( i
α − φ ) ( i
α − φ )⎦
(6.38)

25
 When CMO and constant relative
volatility (i.e. αi = αi ), there are common
values of φ and φ that satisfy both Eqs.
6.36 & 6.37, thus making Eq. 6.38 become
⎡αi Dx i , dist αi Bx i , bot ⎤
Vmin − Vmin = ∑⎢ + ⎥
(
⎣ i α − φ ) ( α i − φ ) ⎦
(6.39)
or
⎡αi ( Dx i , dist + Bx i , bot ) ⎤
Vmin − Vmin = ∑⎢ ⎥
⎣ (α i − φ ) ⎦
(6.40)

The overall column material balance

equation for species i can be written as
follows
Fz i = Dx i , dist + Bx i , bot (6.41)

26
 Combining Eq. 6.41 with Eq. 6.40 yields
αi Fz i
Vmin − Vmin = ΔVfeed =∑
(α i − φ )
(6.42)

Note that ΔVfeed or VF , min is the change in

vapour flow rate at the feed stage

If the value of q is known,

ΔVfeed
=1− q
F
or
ΔVfeed = (1 − q ) F (6.43)

27
 Eq. 6.42 is the first Underwood
equation, use to approximate the value of φ ,
and Eq. 6.36 is the second Underwood
equation, used to compute the value of Vmin

Once Vmin is known, the value of Lmin can

then be calculated from the balance at the
condenser, as follows
Lmin = Vmin − D (6.44)

Note that
D = ∑ ( Dx i , dist ) (6.45)

28
 The use of Underwood equations can be
divided into 3 cases as follows

Case A: Assume all NKs do not

distribute; i.e. for the distillate,
Dx HNK, dist = 0
and
Dx LNK, dist = Fz LNK
while the amounts of key components are
Dx LK, dist = ( FRLK ) Fz LK
Dx HK, dist = (1 − FRHK ) Fz HK (6.46)

Eq. 6.42 can be solved for the value of φ

between the relative volatilities of LK and
HK, or α HK < φ < α LK

29
 Case B: Assume that the distributions
of NKs obtained from the Fenske equation
at total reflux are still valid/applicable at
minimum reflux

Once again, the value of φ is between

the relative volatilities of LK and HK, or
α HK < φ < α LK

Case C: Exact solution (without having

to assume any assumptions)

If there are C species, there will be C

values for φ

30
 Thus, we can have C–1 degree of
freedoms, which yields C–1 equations for
Eq. 6.36, and there are C–1 unknowns (i.e.
Vmin and Dx i , dist for all LNK and HNK)

With the C–1 unknowns and C–1

equations, the value of φi for each species
can be solved as follows
α HNK, 1 < φ1 < α HNK, 2 < φ2 < α HK < ... < α LK < φC-1 < α LNK, 1

31
Example For the distillation problem on
Page 12, determine the minimum reflux
ratio, based on the feed rate of 100 kmol/h

Since it is given that the feed is a

saturated vapour, q = 0 , which results in
ΔVfeed = (1 − q ) F
= (1 − 0 )(100 )
ΔVfeed = 100

Eq. 6.42 becomes

αi Fz i
ΔVfeed = ∑
(α i − φ )
α benz Fz benz α tolFz tol α cumeFz cume
100 = + +
(α benz − φ ) (α tol − φ ) (α cume − φ )

32
 2.25 (100 )( 0.40 ) 1.0 (100 )( 0.30 ) 0.21(100 )( 0.30 )
100 = + +
( 2.25 − φ ) (1.0 − φ ) ( 0.21 − φ )
(6.48)

Since LK = toluene (α = 1.0 ) and HK =

cumeme (α = 0.21) , the value of φ is
between 0.21 and 1.0

Solving Eq. 6.48 gives the value of φ of

0.5454

The next step is to determine the value

of Vmin using Eq. 6.36
αi Dx i , dist
Vmin =∑
(α i − φ )

33
 Since all species (including the LNK or
benzene) are distributed to both distillate and
bottom products, the value of Dx i , dist of each

species can be computed from the following

equation
Dx i , dist = z i F (FR )i , dist (6.49)

It given in the problem statement (see

Page 12) that
• the fraction recovery ( FR ) of toluene in
the distillate (FR )tol, dist is 95% or 0.95

• the fraction recovery ( FR ) of cumeme in

the bottom (FR )cume, bot is 95% or 0.95;

thus, (FR )cume, dist = 1 − 0.95 = 0.05

From the calculations (see Page 16), the

fractional recovery of benzene (LNK in this
Example) or (FR )benz, dist is found be 0.998

34
 Substituting corresponding numerical
values into Eq. 6.49 yields
• Dx benz, dist = ( 0.40 )(100 )( 0.998 ) = 39.9

• Dx tol, dist = ( 0.30 )(100 )( 0.95 ) = 28.5

• Dx cume, dist = ( 0.30 )(100 )( 0.05 ) = 1.5

Thus, the value of Vmin can be computed

as follows
αi Dx i , dist
Vmin =∑
(α i − φ )
α benz Dx benz, dist α tolDx tol, dist α cumeDx cume, dist
= + +
(α benz − φ ) (α tol − φ ) (α cume − φ )

Vmin =
( 2.25 )( 39.9 )
+
(1.0 )( 28.5 )
( 2.25 − 0.5454 ) (1.0 − 0.5454 )
+
( 0.21)(1.5 )
( 0.21 − 0.5454 )
Vmin = 114.4

35
 Since we have learned that

∑ ( Dx i , dist )=D (5.18)

Thus,
D = Dx benz, dist + Dx tol, dist + Dx cume, dist
= 39.9 + 28.5 + 1.5
D = 69.9

Accordingly,
Lmin = Vmin − D = 114.4 − 69.9 = 44.5

Therefore, the minimum reflux ratio

⎛ L ⎞ is 44.5 = 0.64
⎜ ⎟
⎝ D ⎠min 69.9

36
Gilliland Correlation for Number of
Stages at Finite Reflux Ratio

We have studies 2 extreme cases for

multi-component distillation, i.e. total
reflux (proposed by Fenske) and minimum
reflux (proposed by Underwood)

In order to determine the number of

stages for multi-component distillation at
finite reflux ratio, there should be a
correlation that uses the results from both
extreme cases

37
 E.R. Gilliland (1940) established a
technique that empirically relate the
number of stages, N , at finite reflux ratio
⎛ L ⎞ to the minimum number of stages,
⎜ ⎟
⎝D ⎠
N min (at total reflux) and the minimum

⎛ L⎞
reflux ratio ⎜ ⎟
⎝ D ⎠min

In order to develop this correlation,

Gilliland performed a series of accurate
stage-by-stage calculations and found that
there was a correlation between the

function
( N − N min )
and the function
( N + 1)
⎡⎛ L ⎞ ⎛ L ⎞ ⎤
⎢⎜⎝ D ⎟⎠ − ⎜⎝ D ⎟⎠ ⎥
⎣ min ⎦

⎡⎛ L ⎞ ⎤
⎢⎣⎜⎝ D ⎟⎠ + 1⎥⎦

38
 The correlation was later modified by
C.J. Liddle (1968) and could be presented
in the form of chart as follows

39
 The procedure of using the Gilliland’s
correlation is as follows
1) Calculate N min using the Fenske
equation
⎛ L⎞
2) Calculate ⎜ ⎟ using Underwood’s
⎝ D ⎠min
equations (or analytically for a binary
mixture)
L
3) Choose actual , which is normally
D
within the range of 1.05 to 1.5 times
⎛ L⎞
that of ⎜ ⎟
⎝ D ⎠min
(note the number between 1.05 to 1.5
⎛ L⎞
that uses to multiply ⎜ ⎟ is called
⎝ D ⎠min
a multiplier, M )

40
 4) Calculate the abscissa or the value of
⎡⎛ L ⎞ ⎛ L ⎞ ⎤
⎢⎜⎝ D ⎟⎠ − ⎜⎝ D ⎟⎠ ⎥
⎣ min ⎦

⎡⎛ L ⎞ ⎤
⎢⎣⎜⎝ D ⎟⎠ + 1⎥⎦

5) Determine the ordinate value or the

( N − N min )
value of using the
( N + 1)
correlating line
6) Calculate the actual number of
stages, N

Note that the Gilliland’s correlation

should be used only for rough estimates

41
 The optimal feed stage/plate location
can also be estimated

First, the Fenske equation is used to

determine the minimum number of stages,
N min

Then, the optimal feed stage can be

obtained by determining the number of
stages required to go from the feed
concentrations to the distillate
concentrations for the key components
⎡ ⎛ x LK ⎞ ⎤
⎢⎜ x ⎟ ⎥
ln ⎢ ⎝ HK ⎠dist ⎥
⎢ ⎛ z LK ⎞ ⎥
⎢ ⎜z ⎟ ⎥
NF, = ⎣ ⎝ HK ⎠ ⎦ (6.50)
ln α LK-HK
min

42
 Next, by assuming that the relative feed
location is constant as we change from total
reflux to a finite value of reflux ratio, we obtain
the following equation
N F , min N F
= (6.51)
N min N

Alternatively, a probably more accurate

equation (proposed by Kirkbride – in the
Separation Process Technology by J.L.
Humphrey and G.E. Keller II, 1997) is used to

( )
estimate the optimal feed stage N f , as follows

⎛ N f −1 ⎞ ⎧
⎪ B ⎛ z HK ⎞ ⎛ x LK, bot ⎞ ⎫⎪
2

log ⎜ ⎟ = 0.260 log ⎨ ⎜ ⎟ ⎜ ⎟ ⎬

⎝N − Nf ⎠ ⎪⎩ D ⎝ z LK ⎠ ⎝ x HK, dist ⎠ ⎭⎪
(6.52)

Note, once again, that both Eqs. 6.51 &

6.52 should be used only for a first guess for
the optimal feed location

43
 The Gilliland’s correlation can also be
presented in the form of equation as follows
⎡⎛ L ⎞ ⎛ L ⎞ ⎤
⎢⎜⎝ D ⎟⎠ − ⎜⎝ D ⎟⎠ ⎥
(note that x = ⎣ min ⎦ )

⎡⎛ L ⎞ ⎤
⎢⎣⎜⎝ D ⎟⎠ + 1⎥⎦

For 0 ≤ x ≤ 0.01:
( N − N min ) = 1.0 − 18.5715x
(6.53)
( N + 1)
For 0.01 < x < 0.90 :
( N − N min ) = 0.545827 − 0.591422x + 0.002743
( N + 1) x

(6.54)
For 0.90 ≤ x ≤ 1.0 :
( N − N min ) = 0.16595 − 0.16595x
( N + 1)
(6.55)

44
Example Estimate the total number of
equilibrium stages and optimal feed stage
for the Example on Pages 12 & 32 if the
⎛ L⎞
actual reflux ratio ⎜ ⎟ is set at 2
⎝D ⎠

1) Calculate the value of N min

From the Example on Page 12,
N min = 3.8

⎛ L⎞
2) Calculate the value of ⎜ ⎟
⎝ D ⎠min
From the Example on Page 32,
⎛L ⎞ = 0.64
⎜ ⎟
⎝ D ⎠min

45
 L
3) Choose the value of
D
L
It is given that is set as 2
D

4) Calculate the abscissa

The abscissa can be computed as follows
⎡⎛ L ⎞ ⎛ L ⎞ ⎤
⎢⎜⎝ D ⎟⎠ − ⎜⎝ D ⎟⎠ ⎥ [2 − 0.64]
⎣ min ⎦
= = 0.453
⎡⎛ L ⎞ ⎤
+ 1 [2 + 1]
⎢⎣⎜⎝ D ⎟⎠ ⎥⎦

5) Determine the value of ordinate

The ordinate can be read from the
Gilliland’s correlation chart, and it is found
to be
( N − N min ) ≈ 0.27
( N + 1)

46
 ⎡⎛ L ⎞ ⎛ L ⎞ ⎤
⎢⎜⎝ D ⎟⎠ − ⎜⎝ D ⎟⎠ ⎥
Alternatively, since x = ⎣ min ⎦
=
⎡⎛ L ⎞ ⎤
⎢⎣⎜⎝ D ⎟⎠ + 1⎥⎦

0.453, Eq. 6.54 can be used to calculate the

value of
( N − N min )
as follows
( N + 1)
( N − N min ) = 0.545827 − 0.591422 0.453 + 0.002743
( )
( N + 1) 0.453
= 0.284

6) Calculate the value of N

The number of equilibrium stages, N ,
can be computed as follows
( N − N min ) = 0.27
( N + 1)
( N − 3.8 ) = 0.27
( N + 1)
N − 3.8 = 0.27 ( N + 1)

47
 N − 3.8 = 0.27N + 0.27
0.73N = 4.07
N = 5.58

To determine the optimal feed stage for

the case of total reflux, Eq. 6.50 is used

It is given that (see Page 12)

• z LK = z tol = 0.30
• z HK = z cume = 0.30

From the Example on Page 32, we

obtained the following
• Dx tol, dist = 28.5

• Dx cume, dist = 1.5

• D = ∑ ( Dx i , dist ) = 69.9

48
 Thus, the values of x tol, dist and x cume, dist
can be computed as follows
Dx tol, dist 28.5
• x tol, dist = = = 0.408
D 69.9
Dx cume, dist 1.5
• x cume, dist = = = 0.021
D 69.9

Substituting corresponding numerical

values into Eq. 6.50 results in
⎡ ⎛ 0.408 ⎞ ⎤
⎢ ⎜⎝ 0.021 ⎟⎠ ⎥
ln ⎢ ⎥
⎢ ⎛⎜ 0.30 ⎞⎥
⎟
⎢⎣ ⎝ 0.30 ⎠ ⎥⎦
NF, min = = 1.90
1 ⎞
ln ⎛⎜ ⎟
⎝ 0.21 ⎠

49
 Hence, the optimal feed stage for the
L
case of finite reflux ratio ( = 2 in this
D
Example) can be calculated using Eq. 6.51
as follows
N F , min N F
=
N min N
⎛ N F , min ⎞
NF = ⎜ ⎟ N
⎝ N min ⎠
⎛ 1.90 ⎞
=⎜ ⎟ 5.58
⎝ 3.8 ⎠
N F = 2.79 ≈ 3

50
 Gilliland’s Correlation Chart
(from Separation Process Engineering (2nd ed.) by P.C. Wankat, 2007)

51