Total reflux ratio Operating and optimum reflux ratio

In distillation of a binary mixture A and B, the feed
conditions, distillate and bottoms compositions are
usually specified and the number of theoretical trays are
to be calculated. The number of theoretical trays depends
on the operating lines. To fix the operating lines, the
reflux ratio R = Ln/D at the top must be set.
The actual operating reflux ratio lies between the two
limits. To select the proper value of R requires a
complete economic balance on the fixed costs of the
tower and operating costs. By experience, the optimum
reflux ratio has been shown to be between 1.2Rm and
1.5Rm.

One limiting case is total reflux, R = ∞, or D = 0. The

material balance becomes
Vn+1 = Ln
Vn+1yn+1 = Lnxn
Hence, the operating lines of both sections are on the 45o
line, yn+1 = xn.

Total reflux
is an extreme
case, the
number of
theoretical
trays required
is at its
Example D4: Minimum reflux ratio and total reflux
minimum to
ratio
obtain the For the rectification in Exanple D3, where a benzene-
given
toluene feed is being distilled to give a distillate
separation of composition of xD = 0.95 and a bottom product of xB =
xD and xB. 0.10, calculate the following:
However, in (a) Minimum reflux ratio Rm
reality we
(b) Minimum number of theoretical plates at a total
have no product at all, and the twoer diameter is infinite. reflux
74 75
Solution:
(a) First draw the equilibrium line and the q line as we
did in Example D3. The operating line for minimum
reflux ratio is plotted as a dashed line and intersects the
equilibrium line at the same point the q line intersects.
Reading the values of x’ = 0.49 and y’ = 0.702, we have
′ 0.95 0.702
0.539
1 ′ 0.95 0.49
Hence, the minimum reflux ratio is Rm = 1.17.
(b) The theoretical steps are drawn as shown between the
equilibrium line and the 45o line. The minimum number
of theoretical steps is 5.8, which gives 4.8 theoretical
trays plus a reboiler.
76
Special case for rectification
1. Stripping-column distillation.
In some cases the feed is added at the top of the stripping
column because we would like to have bottoms product
only. The feed is usually a saturated liquid at the boiling
point, and the overhead product VD is the vapor rising
from the top plate, which goes to a condenser with no
relux returned to the tower.
The bottoms product W usually has a high concentration
of the less volatile component B. Hence, the column
operates as a stripping tower, with the vapor removing
the more volatile A from the liquid as it flows downward.
Assuming constant molar flow rates, a material balance
of the more volatile component A around the dashed line
gives,
L Wx
y m 1  m xm  W
Vm 1 Vm 1
77
This stripping line is the same as that for a complete VD = 340 kmol/h
tower. It intersects the y = x line at x = xW, and the slope
is constant at Lm/Vm+1. Component A balance gives
FxF = WxW + VDyD
If the feed is saturated liquid, then Lm = F. This is shown 400(0.70) = 60(0.10) + 340yD
in the figure. Starting at xF, the steps are drawn down the yD = 0.806
tower. For a saturated liquid, q=1, the q line is vertical. The
operating line is plotted through the point y = yW = 0.10
If the feed is cold liquid below the boiling point, the q and the intersection of yD = 0.806 with the q line.
line should should be used and q > 1: Lm = qF. Alternatively, the slope of Lm/Vm+1 = 400/340 = 1.176
can be used. Stepping off the trays from the top, 5.3
Example D5: Number of trays in stripping tower theoretical steps or 4.3 theoretical trays plus a reboiler
A liquid feed at the boiling point of 400 kmol/h are needed.
containing 70 mol % benzene (A) and 30 mol % toluene
(B) is fed to a stripping tower at 101.2 kPa pressure. The 2. Enriching-column distillation
bottoms product flow is 60 kmol/h containing only 10 Enriching towers are also used at times, where the feed
mol % A and the rest B. Calculate the kmol/h overhead enters the bottom of the tower as a vapor. The overhead
vapor, its composition, and the number of theoretical distillate is produced in the same way as in a complete
steps required. fractionating tower and is usually quite rich in the more
volatile component A. The liquid bottoms is usually
Solution comparable to the feed in composition, slightly leaner in
F = 400 kmol/h A. If the feed is saturated vapor, the vapor in the tower
xF = 0.70 Vn = F.
W = 60 kmol/h
xW = 0.10
Plot the equilibrium
and diagonal lines.
Overall material
balance gives
F = W + VD
400 = 60 + VD
78 79
 Plate Efficiency 1.3 Local or point efficiency
y'  y' n1
'  n
 To translate ideal plates into actual plates y' en  y' n1
 Applicable to both distillation & absorption where
y'n = concentration of vapor leaving a specific point in
1 Types of plate efficiency plate n
y'n+1= concentration of vapor entering plate n at the
1.1 overall efficiency same location
y'en = concentration of vapor in equilibrium with liquid
number of ideal plates
0  at the same point (x'n)
number of actual plates
 for a specific location in a plate
 for the entire column
 simple but the least fundamental

1.2 Murphree efficiency

y y
M  n* n1
yn  yn1
where
yn = actural concentration of vapor leaving plate n
yn+1 = actural concentration of vapor entering plate n
yn* = concentration of vapor in equilibrium with liquid
concentration xn leaving downpipe from plate n
2. Relationship between efficiencies
 for a single plate
 in reality samples are taken of the liquid 2.1 Murphree & local efficiencies
on the plates, and the vapor compositions are In small columns, good mixing can be achieved so that
determined from a McCabe-Thiele diagram. the concentration is uniform in the tray.
y'n = yn, y'n+1 =yn+1, and y'en = yn*. Therefore, '  M .

80 81
In large columns, incomplete mixing of the liquid occurs
in the tray.  M is the integration of ' over the entire tray.
2.2 Murphree & overall efficiencies
ln1  M ( mV / L  1)
0 
ln mV / L 
where m is the slope of equilibrium line. When
mV/L=1.0 or M 1.0, M = 0.
This relationship depends on the relative slopes of
equilibrium and operating lines.
M < 0 (stripping section)
M > 0 (enriching section)
M  0 for the whole column if the feed plate is near the
middle
3. Use of Murphree efficiency
 Draw an effective equilibrium line.
 Use McCable and Thiele method between the
effective equilibrium line and the operating line.
4. Factors influencing plate efficiency
 Adequate and intimate contact between liquid and
vapor can enhance the efficiency; any excessive
foaming or entrainment, poor distribution, or short-
circuiting, weeping, or dumping of liquid, lowers
the plate efficiency.
 Rate of mass transfer between liquid and vapor.
82
Fractional Distillation using
Enthalpy-Concentration Method
 molal flow rates need not to be constant
 using enthalpy as well as material balances
When the operating lines are plotted with the
equilibrium curve on the x-y diagram, the McCabe-
Thiele step-by-step construction can be used to compute
the number of ideal plates. However, unless Ln and Lm
are constant, the operating lines are curved and can be
plotted only if the change with concentration in these
internal streams is known. Enthalpy balances are
required in the general case to determine the position of
a curved operating line.
The enthalpy-concentration diagram requires the
following data at a constant pressure:
(1) heat capacity of the liquid as a function of
temperature, composition, and pressure.
(2) heat of solution as a function of temperature and
composition.
(3) latent heats of vaporization as a function of
composition and pressure or temperature.
(4) boiling point as a function of pressure, composition
and temperature.
The diagram at a given constant pressure is based on
arbitrary reference states of liquid and temperature,
which is usually taken as the boiling point of the lower-
boiling component A.
83
The saturated liquid enthalpy is Example D6. Enthalpy-concentration plot for
H x  x AC PLA Tb  T0   (1  x A )C PLB Tb  T0   H sol benzene-toluene
(1)
Tb = boiling temperature of mixture at xA Prepare an
T0 = reference temperature enthalpy-
CPLA = liquid heat capacity of A concentration plot
CPLB = liquid heat capacity of B
for benzene-
toluene at 1 atm
Hsol = heat of mixing, usually ignored
pressure.
Equilibrium and
The saturated vapor enthalpy is physical property
H y  y A HVA  C PVA Td  T0   (1  y A )HVB  C PVB Td  T0 
data are given.
(2)
Td = dew temperature of mixture at yA
T0 = reference temperature
CPVA = vapor heat capacity of A Vapor pressure (kPa) Mole fraction of
CPVB = vapor heat capacity of B T (K) benzene at 101.325 kPa
HVA = latent heat of A at the reference Benzene Toluene xA yA
temperature T0 & calculated from that 353.3 101.32 - 1.000 1.000
at the normal boiling temperature TbA. 358.2 116.9 46.0 0.780 0.900
363.2 135.5 54.0 0.581 0.777
HVA  CPLA TbA  T0   HVAb  CPVA TbA  T0  (3) 368.2 155.7 63.3 0.411 0.632
HVB  CPLB TbB  T0   HVBb  CPVB TbB  T0  (4) 373.2 179.2 74.3 0.258 0.456
378.2 204.2 86.0 0.130 0.261
That is, the liquid is first changed from temperature T0 to 383.8 240.0 101.32 0 0
Tb and then evaporated at Tb , the vapor temperature is finally
Component Boiling Heat capacity Latent heat of
changed back from Tb to T0.
point (kJ/kmol K) vaporization
(oC) Liquid Vapor (kJ/kmol)
Benzene (A) 80.1 138.2 96.3 30820
Toluene (B) 110.6 167.5 138.2 33330
84 85
Solution. By the similar procedure,
We choose a reference temperature of T0=80.1oC so that at xA = 0.3 and yA = 0.3,
the liquid enthalpy of pure benzene (xA=1) at the boiling Hx = 2920 kJ/kmol and Hy = 36268 kJ/kmol
point is zero. at xA = 0.8 and yA = 0.8,
Hx = 562 kJ/kmol and Hy = 32380 kJ/kmol
For the first point, we select pure toluene (xA=0). For Enthalpy-concentration data for benzene-toluene mixture at 1 atm.
liquid toluene at the boiling point of 110.6oC, Saturated Liquid Saturated Vapor
H x  x AC PLA Tb  T0   (1  x A )C PLB Tb  T0   H sol Mole Enthalpy, Hx, Mole Enthalpy, Hy,
 0  (1  0)(167.5)(110.6  80.1)  5109 kJ / kmol fraction, xA (kJ/kmol) fraction, yA (kJ/kmol)
0 5109 0 38439
For the saturated vapor enthalpy of pure toluene, we first
0.30 2920 0.30 36268
need to calculate ΔHVB at T0=80.1oC:
0.50 1820 0.50 34716
HVB  C PLB TbB  T0   HVBb  C PVB TbB  T0 
0.80 562 0.80 32380
 167.5(110.6  80.1)  33330  138.2(110.6  80.1) 1.00 0 1.00 30820
 34224 kJ/kmol
Hence, at xA = 0,
H y  y A HVA  C PVA Td  T0   (1  y A )HVB  C PVB Td  T0 
= 0 + (1-0)[34224+138.2(110.6-80.1)]
=38439 kJ/kmol
For pure benzene, xA = yA = 1, T = T0=80.1oC, Hx = 0,
H y  y A HVA  C PVA Td  T0   (1  y A )HVB  C PVB Td  T0 
=1[30820+96.3(80.1-80.1)] + 0 = 30820 kJ/kmol
Selecting xA = 0.5, the boiling point Tb = 92oC and the
dew point for yA = 0.5 is Td = 98.8oC from the T-x-y plot.
H x  x AC PLA Tb  T0   (1  x A )C PLB Tb  T0 
= 0.5(138.2)(92-80.1)+(1-0.5)(167.5)(92-80.1)
= 1820 kJ/kmol
Enthalpy-concentration plot for benzene-toluene mixture at 1 atm.
H y  y A HVA  C PVA Td  T0   (1  y A )HVB  C PVB Td  T0 
The tie line represents the enthalpies and composition of
= 0.5[30820+96.3(98.8-80.1)] the liquid and vapor phases in equilibrium.
+(1-0.5)[34224+138.2(98.8-80.1)]
= 34176 kJ/kmol
86 87
Distillation in Enriching Section of Tower Substitute Eq. (9) into Eq. (8) we have
Material balances Vn 1H y , n 1  Ln H xn  V1H y1  LH xD
Vn 1  Ln  D (5) (10)
 Vn 1  D H xn  V1H y1  LH xD
Vn1 yn 1  Ln xn  Dx D (6)
Eqs. (7) & (10) are the final working equations for the
By rearranging Eq. (6) we obtain the operating line
L Dx enriching section.
yn1  n xn  D (7)
Vn1 Vn1 In order to plot the operating line Eq. (7), the terms Vn+1
Ln and Vn+1 may vary throughout the tower so Eq. (7) and Ln must be determined from Eq. (10). If the reflux
will not be a straight line on an xy plot. ratio is set, V1 and L are know. The values of Hy1 and
HxD can be determined by Eqs. (1) & (2) or from an
enthalpy-concentration plot. If a value of xn is selected,
it is a trial-and-error solution to obtain Hy,n+1 since yn+1
is unknown. The steps to follow are given below.
1. Select a value of xn. Assume Vn+1 = V1 = L+D, then
Ln = Vn+1 - D.
2. Use Eq. (7) to calculate the approximate value of yn+1.
3. Using this yn+1 to obtain Hy,n+1 and xn to obtain Hxn .
Substitute these values into Eq. (10) and solve for the
new Vn+1. Obtain new Ln from Eq. (5)
4. Substitute the above values into Eq. (7) to get the new
yn+1.
5. If the newly calculated yn+1  the assumed value,
repeat steps 3-4. Usually one iteration is enough.
Making an enthalpy balance, Assume another xn and repeat steps 2-5.
Vn1 H y,n1  Ln H xn  DH xD  qc (8) 6. Plot the curved operating line for the enriching section.
where qc is the condenser duty, kJ/h.
An enthalpy balance for the condenser only gives
qc  V1 H y1  LH xD  DH xD (9)
88 89
Distillation in Stripping Section of Tower Example D7. Distillation Using Enthalpy-
Concentration Method
Material balances
Lm  Vm 1  B (11) A liquid mixture of benzene-toluene is to be distilled in a
Lm x m  Vm 1 ym 1  Bx B (12) fractionating tower at 101.3 kPa pressure. The feed of
so that the operating line is 100 kmol/h is liquid, containing 45 mol % benzene and
L Bx 55 mol % toluene, and enters at 54.4oC. A distillate
ym1 m x m  B (13) containing 95 mol % benzene and 5 mol % toluene and a
Vm1 Vm 1
bottoms containing 10 mol % benzene and 90 mol %
Making an enthalpy balance, toluene are to be obtained. The reflux ratio is 1.755. The
Vm 1H y,m1  Vm 1  B H xm  BH xB  qr (14) average heat capacity of the feed is 159 kJ/(kmol K) and
the average latent heat is 32099 kJ/kmol. Use enthalpy
where qr is the reboiler duty, kJ/h. balances to calculate the flow rates of the liquid and
vapor at various points in the tower and plot the curved
An enthalpy balance for the whole distiller gives operating lines. Determine the number of theoretical
qr  DH xD  BH xB  qc  FH F (15) stages needed.
The final working equations are Eqs. (13) and (15).
Solution:
Using a method similar to that of the enriching section to F = 100 kg mol/h, xF = 0.45, xD = 0.95, xB = 0.10,
solve the equations. R = 1.5Rm = 1.5(1.17) = 1.755,
D = 41.2 kg mol/h, B = 58.8 kmol/h
CP (kJ/(kmol K) ΔHvb The feed enters at 54.5oC so q = 1.195.
o
Component Tb ( C) Liquid Vapor (kJ/kmol) The flows at the top of the tower are calculated as
Benzene 80.1 138.2 96.3 30820 follows.
Toluene 110.6 167.5 138.2 33330 R = L/D = 1.755; L = 1.755D = 1.755(41.2) = 72.3;
V1 = L+D = 72.3+41.2 = 113.5

The latent heats of vaporization of benzene and toluene

at the reference temperature of 80.1oC are
HVA = HVAb = 30820 kJ/kmol
90 91
HVB  CPLB TbB  T0   HVBb  CPVB TbB  T0  This calculated value is close enough to the approximate
= 167.5(110.6-80.1) + 33330 - 138.2(110.6-80.1) value of 0.695 so that no further trials are needed.
= 34224 kJ/kmol
Select other values of xn and calculate the responding
The saturation temperature (dew point) at the top of the yn+1, plot the curved operating line in the enriching
tower for y1 = xD = 0.95 is 82.3oC. section.
Hy1 = 0.95[30820+96.3(82.3-80.1)]
The condenser duty is obtained from Eq. (9)
+(1-0.95)[34224+138.2(82.3-80.1)] = 31206 kJ/kmol qc  V1 H y1  LH xD  DH xD
 113. 5(31206)  72. 3(139)  41. 2(139)  3526100 kJ / h
The boiling point of the distillate D is obtained as
For xB = 0.10, HxB = 4350 from the Figure in page 89.
81.1oC.
The feed is at 54.5oC, using Eq. (1), we have
HxD = 0.95(138.2)(81.1-80.1)
HF = 0.45(138.2)(54.5-80.1)
+ (1-0.95)(167.5)(81.1-80.1) = 139 kJ/kmol
+ (1-0.45)(167.5)(54.5-80.1) = -3929 kJ/kg mol
Using Eq. (15),
Step 1: select xn = 0.55 and guess yn+1 from Eq. (7)
Ln Dx D 72. 3 41. 2 qr = 41.2(139) + 58.8(4350) + 3526100 - 100(-3929)
yn1  xn   ( 0. 55)  ( 0. 95)  0. 695 = 4180500 kJ/h
Vn1 Vn1 113. 5 113. 5
Making a material balance below the bottom tray and
Step 2: using the Figure in Example D6 for xn = 0.55
around the reboiler we have
obtain Hxn = 1590 and for yn+1 = 0.695, Hy,n+1 = 33240.
LN  B  Vb (16)
Substituting into Eq. (10)
Enthalpy balance
Vn+1 (33240) = (Vn+1 - 41.2)1590
Vb H yb  Vb  B H xN  BH xB  qr (17)
+113.5(31206) - 72.3(139)
Vn+1 =109.5 From the equilibrium diagram we find that for xB = 0.10,
yB = 0.207, which is the vapor composition leaving the
Using Eq. (5),
reboiler.
109.5=Ln +41.2 or Ln =68.3
For the equimolal overflow in the stripping section,
Step 3: substituting into Eq. (7),
Lm  Ln  qF  72. 3  1.195(100 )  191. 8 (18)
68. 3 41. 2
yn1  ( 0. 55)  ( 0. 95)  0. 700 Vm 1  Vn 1  (1  q ) F  113. 5  (1  1.195)100  133. 0
109. 5 109. 5
92 93
Selecting ym+1 = yB = 0.207 and using Eq. (13), an
approximate value of xm = xN is obtained.
L Bx
ym1 m x m  B
Vm1 Vm 1
191. 8 58. 8
0. 207  xN  ( 0.10 )
133. 0 133. 0
Solving xN =0.174. From the Figure in example D6 for
xN =0.174, HxN =3800, and for yB = 0.207, Hyb =37000.

Substituting into Eq. (17),

Vb (37000)  Vb  58.83800  58.8(4350)  4180500
Solving Vb = 125.0. Using Eq. (16) we get LN = 183.8.

Substituting into Eq. (13) and solving for xN.

183. 8 58. 8
0. 207  xN  ( 0.10 )
125. 0 125. 0 Solid line: enthalpy-concentration method
xN = 0.173 Dashed line: constant molal overflow
This value is quite close to the approximate value of The solid line is slightly above the dashed line.
0.174.

Assuming other values of ym+1 and solving for xm using

the same procedure, the curved operating line for the
stripping section can be constructed.

The number of theoretical stages is determined to be

10.4, compared with 9.9 steps using the constant flow
method.

94 95
Ponchon and Savarit method This can be explained by the lever rule. A vapor of Vn+1
of composition y n+1 is separated into a distillate D of xD,
Rectifying section and a liquid Ln of xn. (Vn+1, yn+1) is at the fulcrum of a
We rearrange Eq. (8) as beam, balanced by the distillate (D, xD) and liquid (Ln,
Vn1H y,n1  Ln H xn  DH xD  qc  DQ' (16) xn). In order for the beam to be “balance”, the moments
DH xD  qc about the fulcrum must be equal. Hence, we have Eq.
where Q'  (17).
D A similar analysis for Eq. (16) gives
The left hand side is the net rate flow of heat upwards
through the enriching section, while the right hand side
Ln  D H y,n1  Ln H xn  DQ'
is constant. This means that the net rate of flow of heat  
Ln H y,n1  H xn  D Q' H y ,n1  
is constant and independent of the tray number in the or
rectifying section. Ln 
Q' H y ,n 1 
  internal reflux ratio (19)
Using Eq. (5) to eliminate Vn+1 in Eq. (6), we have
D 
H H
y ,n 1 xn 
Ln  D  yn 1  Ln xn  DxD Again this obeys the lever rule.
Ln  y n 1  xn   D  x D  y n 1  (17)
Hence,
Ln  x D  y n 1 
  internal reflux ratio (18)
D  y n 1  x n 

Combining Eqs. (18) & (19), we have

Ln  x D  y n 1  Q' H y,n1 
 
D  y n 1  x n  H y ,n 1  H xn  (20)
On the H-x-y diagram, this is a straight line through the
points (Hy,n+1, yn+1) at Vn+1, (Hx,n, xn) at Ln, and (Q’, xD)
at ΔD as linked above and shown in the figure below. ΔD
is termed the difference point.
96 97
 Equilibrium data are used to determine the liquid and
vapor compositions leaving the first theoretical plate.
The graphical solution of Ponchon and Savarit is
continued using the equilibrium data and the H-x-y
diagram until the reboiler composition is reached. In the
following case, a total of 4 theoretical plates is shown, 3
in the column and 1 for the reboiler having an enthalpy
HR and composition xR.

Since the heat and mass balances apply to any tray, then
at the top plate, n = 0, we have
L  x  y1  Q ' H y ,1  line length Q ' H y ,1
R 0  D  
D  y1  x D  H y ,1  H xD  line length H y ,1  H xD

The line length H y ,1  H xD is the latent heat of

vaporization at a distillate composition xD. If the reflux
ratio is specified, then the length of Q’ – Hy,1 can be
calculated, and hence Q’ determined.

98 99
Stripping section tower) to ΔB, intersects the saturated liquid enthalpy
The same analytical techniques used in the rectifying curve at LZ, the liquid leaving the bottom plate. Vapor
section can be applied to the stripping section and the VZ leaving the bottom plate is in equilibrium with the
entire column. liquid LZ and is located on the tie line Z.

The number of theoretical plates in the stripping section

can be determined from the H-x-y diagram by
alternatively constructing lines to ΔB and tie lines. Each
tie line is a theoretical plate.

The complete column

We rearrange Eq. (14 ) as

Lm H xm  Vm1H y,m1  BH xB  qr  BQ' '
where Q” = HxB – qr/B, by eliminating B (B=Lm – Vm+1),
Lm  y  x  H y ,m1  Q"
 m 1 B 
Vm 1  xm  xB  H x,m  Q" (21)
On the H-x-y diagram, this is a straight line passing
throught (Hy,m+1, ym+1) at Vm+1, (Hx,m, xm) at Lm, and (Q”,
xB) at ΔB as shown in the figure above. ΔB is also termed
the difference point. B is W in the figure.

Eq. (21) applies to every plate in the stripping section.

On the H-x-y diagram, this is a line from VZ+1 (vapor
leaving the reboiler and entering the bottom tray Z of the
100 101
The overall mass balance is
F=D+W
The more volatile component mass balance is
FxF = DxD + WxW
The heat balance is
FHF = QC + DHyD + WHxW – QR = DQ’+WQ”
Eliminating F yields
D  x F  xW   H F  Q"
 
W  x D  x F  Q ' H F 
This is a straight line on the H-x-y diagram passing
through (Q’, xD) at ΔD, (HF, xF) at F, and (Q”, xW) at ΔW
and is shown in the above figure for a case of liquid feed
below its boiling point (cold liquid).
The procedure to determine the number of theoretical
plates is summarized below.
1. Locate the feed enthalpy and composition
2. Locate xD and xW
3. Locate ΔD by computation of Q’ or for a specific
reflux ratio as R = L0/D
4. The line ΔD, F is extended to x = xW and thus locates
ΔW whose coordinates can be used to calculate QR.
5. The number of theoretical plates is then determined
by application of the straight line relationship on the
H-x-y diagram and the equilibrium data of the y-x
diagram.
102
Example D8. Distillation Using Ponchon Savarit
Method
Re-calculate the number of theoretical stages required
for example D7 using Ponchon Savarit method.
xF = 0.45, HF = -3929 kJ/kg mol
xD = 0.95, xB = 0.10
Q' 
DH xD  qc 41.2(139)  3526100
  85724 kJ/kmol
D 41.2
Q” = HxB – qr/B =435 -4180500/58.8 = -66747 kJ/kmol
or in the H-x-y diagram, draw the line from ΔD (xD=0.95,
Q’=85724) through (xF=0.45, HF=-3929) to intersect
xB=0.10 gives Q” = -66747 kJ/kmol from ΔB.
The number of theoretical plates is then determined by
application of the straight line relationship on the H-x-y
diagram and the equilibrium data of the y-x diagram.
10.4 plates, or 9.4 plates plus 1 reboiler are needed. The
feed is at plate 6.
103
104 105
 Multicomponent Distillation
 need more than one distillation tower, for n
components, n-1 fractionators are required
 Specification Limitations
The following are established at the beginning
1. Temperature, pressure, composition, and rate of
the feed.
2. Pressure of the distillation
3. Feed to be introduced on that tray which will
result in the least total number of trays
(optimum feed-tray location)
4. Heat loses (even if assumed to be zero)
Under these circumstances only three additional items of
the following list need to be specified, any other items
must be calculated.
1. Total number of trays.
2. Reflux ratio
3. Reboil ratio (ratio of vapor produced by the
reboiler to residue withdrawn)
4. Concentration of one component in one product
(a maximum of two may be chosen)
5. Ratio of flow of one component in the distillate
to the flow of the same component in the
residue, or split of the component (a maximum
of two may be chosen)
6. Ratio of total distillate to total residue
106
 Key components
The feed components are listed in order of their relative
volatility. The more volatile components are called light,
the less volatile are heavy. The components separated in
a tower are called light key (more volatile) and heavy key
(less volatile). Any components lighter than the light
key are called light components, while those heavier than
the heavy key are called heavy components.
In hydrocarbon systems, because of nonidealities, the
equilibrium data are often represented by
yA = KAxA, yB= KBxB, yC = KCxC, yD = KDxD
where KA is the vapor-liquid equilibrium constant or
distribution coefficient for component A. K is a function
of temperature and pressure, but is assumed to be
independent of composition, which is sufficiently
accurate for most engineering calculations. For an ideal
system, KA = PA*/PT = vapor pressure/total pressure.
The relative volatility in multicomponent distillation is
always computed with respect to the heavy key
K y /x
 j  j  je je (1)
K hk yhke / x hke
The value of  is greater than 1 for any component
lighter than the heavy key and less than 1 for any heavier
component.
107
 Bubble-point and dew-point calculations Component Pi* at 105oC, Ki = Pi*/1.2 xi yi=Kixi
These are required for a flash distillation and for each atm
stage of a multicomponent distillation, calculated by 1. Hexane 2.68 2.23 0.33 0.7359
trial-and -error. 2. Heptane 1.21 1.01 0.37 0.3737
NC NC
For the bubble point,  yi   Ki xi 1. 0 (2) 3. Octane 0.554 0.462 0.30 0.1386
i 1 i 1 ∑ = 1.248
NC NC yi
For the dew point,  xi   1. 0 (3)
i 1 i 1 Ki Since ∑yi is too large, try a lower temperature. Since the
First a temperature is assumed and the values of Ki are major contribution is from the first term, try a
obtained from literature. If the summation  1 choose temperature where Ki is lower by a factor of 1/1.24,
another temperature. 2.68/1.24=2.16. Choose T = 96oC, where P1* = 2.16 atm.

Example D9. Find the Component Pi* at Ki = xi Kixi yi

bubble-point and the 96oC, atm Pi*/1.2
dew-point 1. Hexane 2.16 1.8 0.33 0.5940 0.604
temperatures and the 2. Heptane 0.93 0.775 0.37 0.2868 0.292
corresponding vapor 3. Octane 0.41 0.342 0.30 0.1025 0.104
and liquid ∑ = 0.9833 1.000
compositions for a
mixture of 33 mole % This is a liitle bit small. Try another T = 97oC, close
n-hexane 37 mole % enough to 96oC so that y .
∑
n-heptane and 30
mole % n-octane at 1.2
atm total pressure. Dew point. The dew point is higher than the bubble point,
Solution. The logP vs so use 105oC as a first guess.
T plot is shown as Component Ki yi yi / Ki
right. 1. Hexane 2.23 0.33 0.148
2. Heptane 1.01 0.37 0.366
Bubble point. Guess T = 105oC, this is based on the 3. Octane 0.462 0.30 0.649
vapor pressure of heptane, the middle component, is 1.2 ∑ = 1.163
atm.
108 109
Since the sum is too high, try a higher temperature, T = Example D10. The mixture of example D9 is subjected
110oC. to a flash distillation at 1.2 atm pressure, and 60% of the
Component Pi* at Ki = yi yi /Ki xi feed is vaporized.
o
110 C Pi*/1.2 (a) Find the temperature of the flash and the
1. Hexane 3 2.5 0.33 0.132 0.130 composition of the liquid and vapor products.
2. Heptane 1.38 1.15 0.37 0.3217 0.317 (b) To what temperature must the feed liquid be
3. Octane 0.64 0.533 0.30 0.5625 0.553 heated for 60% vaporization on flashing?
∑ = 1.0162 1.000 Solution.
By extrapolation, the dew point is 110.5oC, and the (a) The flash temperature must lie between the bubble
composition of the liquid in equilibrium with the vapor point (97oC) and the dew point (110.5oC).
is calculated by x
/ /
. Assume T = 97+0.6(110.5-97) = 105 oC.
∑ / . From the figure in example D9, K1 =2.68/1.2=2.23,
K2 =1.21/1.2=1.01, K3 =0.554 /1.2=0.462.
 Flash distillation The value of f is 0.6.

Material balance for component i gives The right hand side of Eq. (6) becomes
f 1 x . . .
yDi  x Bi  Fi (4)
f f . . . . . .
where f=V/F is the fraction of the feed vaporized, yDi is =0.190+0.368+0.443=1.001
in equilibrium with xBi. So yDi  Ki x Bi , substitute this
into Eq. (4) and solve for xBi. This is close to unity so the flash temperature is 105 oC.
x Fi The composition of the liquid product is 19% n-hexane,
x Bi  (5) 36.8% n -heptane and 44.2% n -octane.
f  K i  1  1
The summation on xBi should be unity. The composition of the vapor product is computed from
NC NC xFi
 xBi  1   (6) the values of K and x.
i 1 i 1 f  K i  1  1 n-hexane, y = 0.190(2.23) = 0.424
This equation is again solved by iteration. The final n –heptane, y = 0.368(1.01) = 0.372
values of T and Ki are used to calculate the composition n –octane, y = 0.443(0.462) = 0.204
of the top and bottom product streams. 1.000

110 111
(b) To determine the temperature of the feed before  Total Reflux for multicomponent distillation
flashing, an enthalpy balance is made using 105 oC as the (Minimum stages for total reflux)
reference temperature. The heats of vaporization at 105
o
C and the average heat capacities of the liquid from 105 For binary distillation, if the relative volatility of the two
to 200 oC are obtained from the literature. components is approximately constant, we have the
Cp, cal/(mol K) ΔHv, cal/mol following Fenske equation to calculate the minimum
n-hexane 62 6370 number of theoretical steps when a total condenser and a
n –heptane 70 7510 reboiler are used.
n –octane 78 8560  x 1  xB 
ln D 
1 
Nm   B 
x D x
Based on liquid at 105 oC, the enthalpies of the product 1 (7)
ln  AB
are
Hvapor = 0.6(0.424x6370+0.372x7510+0.204x8560) Similarly for any two components (i & j) in
= 4345 cal multicomponent distillation,
Hliquid = 0 x / x 
ln Di Bi 
 x Dj / x Bj 
For the feed, Nm  1 (8)
ln  ij
C 0.33x62 0.37x70 0.30x78
 ij  3  Dij Fij Bij (9)
69.8 cal/ mol °C
Hence, 69.8(T0 – 105) = 4345 The subscripts D, F and B in Eq. (9) refer to the
T0 = 167oC = preheat temperature temperatures of the distillate, feed plate, and bottoms in
the column. Nm is the minimum number of equilibrium
stages excluding the reboiler, at total reflux.

The Fenske equation is derived as below. At total reflux,

D = B = 0, so V = L. Consider the equilibrium partial
reboiler for any two components A and B,
y /x
α
y /x
Rearrange to obtain
112 113
 If we define αAB as the geometric average relative
α (10)
volatility,
Material balances for these two components around the α α α …α α α / (18)
reboiler are Eq. (17) becomes
VRyA,R = LNxA,N (11a)
VRyB,R = LNxB,N (11b) α (19)
Since VR = LN for total reflux (B = 0), we have Solving Eq. (19) for Nm, we have
yA,R = xA,N, yB,R = xB,N (at total reflux) (12) /
Combining Eqs. (10) and (12), /
N 1 (20)
α (13)
or
If we move up the column to stage N, the equilibrium is /
/
α (14) N 1 (21)

The mass balances around stage N simplify to

yA,N = xA,N-1 and yB,N = xB,N-1 The fractional recovery of A in the distillate is
Combining these equations to give FR (22)
α (15) The fractional recovery of A in the bottoms is
Eqs. (13) and (15) can be combined to give FR 1 FR (23)
α α (16) Hence
Dx FR Fx (24)
Which relates the ratio of liquid mole fractions leaving Bx 1 FR Fx (25)
stage N-1 to the ratio in the reboiler.
The Fenske equation can also be written as
Repeating this procedure until we reach the top stage, we /
have /
N 1 (26)
α α …α α α (17)

114 115
Example D11. A mixture with 33% n-hexane, 37% n- the flash temperature were taken from the previous
heptane, and 30% n-octane is to be distilled to give a example.
distillate product with 0.01 mole fraction n-heptane and  Lk,HK  2.23 / 1.01  2.21
a bottoms product with 0.01 mole fraction n-hexane.  x D,LK / x B,LK 
The column will operate at 1.2 atm. 60% of the feed is ln   0.99 / 0.01 
ln  
vaporized. Calculate the complete product compositions x /
 D,HK B,HK  x  0.01 / 0.544 
Nm  1  1
and the minimum number of ideal plates at infinite ln  LK , HK ln 2.21
reflux.  10.8  1  9.8
So the minimum number of ideal stages is 9.8 plus a
Solution reboiler.
light key (LK) heavy key (HK) heavy nonkey (HNK)
n-hexane n-heptane n-octane A more accurate estimate of Nm can be obtained using a
Assume no n-octane and 0.99 mole fraction n-hexane in mean relative volatility based on the values at the top,
the distillate. Take 100 mol/h of feed as the basis middle, and bottom of the column. The top temperature
overall M.B. F = D + B = 100 is about 75oC, the boiling point of n-hexane at 1.2 atm,
hexane M.B. FxF = DxD + BxB and the relative volatility is 2.53. The bottom
o
100x0.33 = 0.99D+(100-D)(0.01) temperature is about 115 C by using a bubble point
D=32/0.98=32.65 mol/h; B=100-D = 67.35 mol/h calculation for the bottoms product, giving a relative
The composition of the bottoms product can be volatility of 2.15.
calculated since this stream contains all the octane, all  LK , HK  3 2.53 2.21 2.15  2.29
but 0.01D of the heptane, 0.01 mole fraction of hexane. Hence, Nm =9.4
FxF DxD xD BxB xB K at 105oC,
1.2 atm
To check our assumption of no octane in the distillate,
LK n-C6 33 32.32 0.99 0.68 0.010 2.23
the Fenske equation can be applied to heptane and
HK n-C7 37 0.33 0.01 36.67 0.544 1.01
HNK n-C8 30 0 0 30 0.446 0.462 octane using  = K2/K3 = 1.01/0.462 = 2.19:
D=32.65 B=67.35  0.01 / 0.544 
ln 
Use the Fenske equation to obtain the minimum number  x D3 / 0.446 
of plates. The relative volatility of the light key to the N m  1  10.4 
ln 2.19
heavy key is the ratio of their K factors. The K values at from which xD3 = 2.4x10-6, which is negligible.

116 117
 Minimum Reflux Ratio for 2. Underwood method
multicomponent distillation By using constant average ’s and assuming
constant molal overflow (L/V), the following
At this reflux ratio, the desired separation is just two equations are solved to obtain RDm:
 i x Fi
barely possible, but an infinite number of plates is 1 q   (28)
required. The desired separation is usually refers to i  
Vmin x
the amount of light key recovered in the distillate  R Dm  1   i Di (29)
and the amount of heavy key recovered in the D i  
bottoms. where q is the moles of liquid to stripping section
of column per mole of feed,  is a parameter to be
Since the only purpose of obtaining the minimum found from Eq. (28). The correct value of  should
reflux ratio is to estimate the product compositions lie between the values of  for the keys.
at RDm and to ensure that the specified RD is
reasonable, an exact value is not required. There
are two approximation methods in estimating RDm:

1. Treating the mixture as a pseudobinary

(light & heavy keys)
L min  Dx D,LK / Fx F,LK    LK,HK  Dx D,HK / Fx F,HK 

F  LK,HK  1
(27)

118 119
Example D12. A mixture with 4% n-pentane, 40% (a) By using the pseudobinary method, we have
n-hexane, 50% n-heptane, and 6% n-octane is to be L min 39.2 / 40  2.48(0.5 / 50)
  0.645
distilled at 1 atm with 98% of the hexane and 1% of F 2.48  1
the heptane recovered in the distillate. What is the L min L min F 1
minimum reflux ratio for a liquid feed at the boiling R Dm    0.645  1.48
D F D 0.437
point?
(b) By using the Underwood method, the K values at 80
o
Solution C are converted to relative volatilities and the  value
The light key is n-hexane and the heavy key is n- between 1 and 2.48 is found by trial. Since q=1 (feed at
heptane. The other components are sufficiently the boiling point), we should have
different in volatility to be undistributed. The x
 i Fi  1  q  1  1  0
known information is summarized below (assume i  
100 moles of feed).
i xFi  i x Fi  i x Fi
 , =1.5  , =1.48
xF FxF DxD xD BxB xB i   i  
n-C5 0.04 4 4 0.092 0 0 n-C5 6.46 0.040.052 0.052
LK n-C6 0.40 40 39.2 0.897 0.8 0.014 n-C6 2.48 0.401.012 0.992
HK n-C7 0.50 50 0.5 0.011 49.5 0.879 n-C7 1.0 0.50-1.00 -1.042
n-C8 0.06 6 0 0 6 0.107 n-C8 0.41 0.06
D=43.7 B=56.3 -0.023 -0.023
0.041 -0.021
K80oC KxF The bubble point of the feed is By further trials or interpolation, =1.487, so
n-C5 3.62 0.145 obtained by checking the
n-C6 1.39 0.556  i x Di 6.64(0.092) 2.48(0.897) 1(0.011)
NC NC R Dm  1     
n-C7 0.56 0.280 equation  y Fi   K i x Fi 1.0 , this  i   6.64  1.487 2.48  1.487 1  1.487
n-C8 0.23 0.014 i 1 i 1
temperature is found to be 80 oC. = 0.119 + 2.24 - 0.023 = 2.336
The corresponding LK, HK is 1.39/0.56 = 2.48.
RDm = 1.34, which is 10% less than the value
obtained using pseudobinary method.

120 121
 Number of ideal plates at operating reflux Example D13. Estimate the number of ideal plates
(Gilliland correlation) required for the separation specified in Example
D11, if the reflux ratio is 1.5RDm.
Gilliland proposed a simple empirical method to
estimate the number of plates in multicomponent Solution
distillation. The correlation requires knowledge From Example D11, the minimum number of ideal
only of the minimum number of plates at total stages is 9.4 plus a reboiler, or 10.4. The value of
reflux and the minimum reflux ratio. The RDm is obtained by the Underwood method.
correlation is given below. xF xD K 
LK n-hexane 0.33 0.99 2.23 2.21
HK n-heptane 0.37 0.01 1.01 1.0
n-octane 0.30 0 0.462 0.457
For a saturated liquid feed, q=1,
 i x Fi
  1 q  11  0
i  
By trial, =1.45,
 i x Di 2.21(0.99) 1.0(0.01)
R Dm  1      2.86
 i   2.21  1.45 1  1.45
RDm = 1.86, RD = 1.5x1.86 = 2.79
R D  R Dm 2.79  1.86
  0.245
RD 1 2.79  1
From the Gilliland correlation,
N  N min
 0.41
N 1
N - 10.4 = 0.41N + 0.41
N is the total number of plates in the system, N = 10.81/0.59 = 18.3 stages, including the reboiler
including the reboiler. The same for Nmin.
122 123