Vous êtes sur la page 1sur 208

Separation Process Technology

Course by Prof. Marco Mazzotti


Summary by Michael Ehrenstein
December 23, 2014

Contents
1 Fundamentals
1.1 Liquid-vapour equilibrium (LVE) . . . . . . . . . . . . .
1.1.1 General expression for the iso-fugacity condition
1.2 Binary systems representation . . . . . . . . . . . . . . .
1.3 Gibbs Phase Rule . . . . . . . . . . . . . . . . . . . . . .
1.3.1 Single component systems . . . . . . . . . . . . .
1.3.2 Binary systems . . . . . . . . . . . . . . . . . . .
1.4 Raoults law (ideal systems) . . . . . . . . . . . . . . . .
1.5 Distribution coefficients . . . . . . . . . . . . . . . . . .
1.6 Real systems and Raoults law . . . . . . . . . . . . . .
1.7 Constant of relative volatility (CRV) . . . . . . . . . . .
1.7.1 Binary systems . . . . . . . . . . . . . . . . . . .
1.7.2 Multicomponent systems . . . . . . . . . . . . .
1.8 Bubble and dew point . . . . . . . . . . . . . . . . . . .
1.8.1 Bubble point . . . . . . . . . . . . . . . . . . . .
1.8.2 Dew point . . . . . . . . . . . . . . . . . . . . . .
1.9 Henrys law . . . . . . . . . . . . . . . . . . . . . . . . .
1.10 Ternary systems representation . . . . . . . . . . . . . .
1.11 Azeotropic systems . . . . . . . . . . . . . . . . . . . . .
1.12 Fundamental principles of mass transfer . . . . . . . . .
1.13 Ficks law . . . . . . . . . . . . . . . . . . . . . . . . . .
1.14 Film model . . . . . . . . . . . . . . . . . . . . . . . . .
1.14.1 Steady state ordinary molecular diffusion . . . .
1.14.2 One-film model . . . . . . . . . . . . . . . . . . .
1.14.3 Two-film model . . . . . . . . . . . . . . . . . . .
1.14.4 The overall mass transfer coefficient (MTC) . . .
1.15 Numerical methods - principals . . . . . . . . . . . . . .
1.16 Half interval method . . . . . . . . . . . . . . . . . . . .
1.17 Newton method . . . . . . . . . . . . . . . . . . . . . . .
1.18 Lever-arm rule . . . . . . . . . . . . . . . . . . . . . . .
2 Contactors
2.1 Gas-liquid contactors . . . .
2.1.1 Principles . . . . . .
2.1.2 Equipment overview
2.2 Tray columns . . . . . . . .
2.2.1 Description . . . . .
2.2.2 Tray types . . . . .
2.2.3 Operative conditions
2.3 Packed columns . . . . . . .
2.3.1 Description . . . . .
2.3.2 Packing types . . . .

.
.
.
.
.
.
.
.
.
.
2

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

6
6
6
7
9
9
10
10
11
12
13
13
14
14
14
15
15
16
19
22
23
23
23
24
25
26
27
27
28
29

.
.
.
.
.
.
.
.
.
.

30
30
30
30
31
31
33
35
36
36
37

2.4

2.3.3 Comparison between packed columns and tray columns 38


Liquid-liquid contactors . . . . . . . . . . . . . . . . . . . . . 39

3 Flash distillation
3.1 Introduction . . . . . . . . . . . . . . . . . . . . .
3.1.1 Definitions and aims . . . . . . . . . . . .
3.1.2 Process description . . . . . . . . . . . . .
3.2 Binary flash . . . . . . . . . . . . . . . . . . . . .
3.2.1 Problem description . . . . . . . . . . . .
3.2.2 Data, specifications, unknowns . . . . . .
3.2.3 Process design . . . . . . . . . . . . . . .
3.2.4 Process verification . . . . . . . . . . . . .
3.3 Multicomponent flash . . . . . . . . . . . . . . .
3.3.1 Problem description . . . . . . . . . . . .
3.3.2 Data, specifications, unknowns . . . . . .
3.3.3 Split factor . . . . . . . . . . . . . . . . .
3.3.4 Bubble point . . . . . . . . . . . . . . . .
3.3.5 Dew point . . . . . . . . . . . . . . . . . .
3.4 Nonideal flash . . . . . . . . . . . . . . . . . . . .
3.4.1 Process design . . . . . . . . . . . . . . .
3.5 Flash cascades . . . . . . . . . . . . . . . . . . .
3.5.1 Flash operations at constant temperature
3.5.2 Flash operations at constant pressure . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

39
39
39
40
40
40
41
42
45
45
45
45
48
49
49
49
49
51
51
52

4 Absorption and stripping


4.1 Introduction . . . . . . . . . . . . . . . . . . .
4.1.1 Definitions and aims . . . . . . . . . .
4.1.2 Process description . . . . . . . . . . .
4.2 Gas plant . . . . . . . . . . . . . . . . . . . .
4.2.1 Solvent selection . . . . . . . . . . . .
4.2.2 Industrial schemes . . . . . . . . . . .
4.3 Absorption in a single equilibrium stage . . .
4.3.1 Assumptions . . . . . . . . . . . . . .
4.3.2 The liquid-vapour equilibrium . . . . .
4.3.3 Single equilibrium stage . . . . . . . .
4.3.4 Co-current cascade configuration . . .
4.3.5 Cross-current cascade configuration .
4.3.6 Counter-current cascade configuration
4.4 Absorption operations in tray columns . . . .
4.4.1 Mass balance . . . . . . . . . . . . . .
4.4.2 The operating line . . . . . . . . . . .
4.4.3 Problem description . . . . . . . . . .
4.4.4 Process design and verification . . . .
4.4.5 Process design - linear case . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

54
54
54
55
56
56
57
59
59
60
62
66
70
74
79
79
80
82
83
85

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

4.4.6 Process design - nonlinear case . . . . . . . . . .


4.4.7 Process verification - linear case . . . . . . . . .
4.4.8 Process verification - nonlinear case . . . . . . .
4.5 Stripping operations in tray columns . . . . . . . . . . .
4.5.1 Comparison: absorption vs. stripping . . . . . .
4.5.2 Process design . . . . . . . . . . . . . . . . . . .
4.5.3 Process verification . . . . . . . . . . . . . . . . .
4.6 Nonideal operations . . . . . . . . . . . . . . . . . . . .
4.6.1 Absorption with efficiency stage . . . . . . . . .
4.7 Multicomponent operations . . . . . . . . . . . . . . . .
4.7.1 Design of a multisolute absorption . . . . . . . .
4.8 Non-isothermal operations . . . . . . . . . . . . . . . . .
4.8.1 Design of a non-isothermal operation . . . . . . .
4.9 Absorption operations in packed columns . . . . . . . .
4.9.1 Problem description . . . . . . . . . . . . . . . .
4.9.2 Design based on the HTU/NTU concept . . . . .
4.10 Chemical absorption . . . . . . . . . . . . . . . . . . . .
4.10.1 Principles . . . . . . . . . . . . . . . . . . . . . .
4.10.2 Case study: H2 S absorption with Diethanolamine

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

88
91
94
97
97
98
101
102
103
107
107
111
111
119
119
120
127
127
128

5 Distillation
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1.1 Definitions and aims . . . . . . . . . . . . . . . . . .
5.1.2 From flash... . . . . . . . . . . . . . . . . . . . . . .
5.1.3 ... to distillation . . . . . . . . . . . . . . . . . . . .
5.1.4 Industrial distillation . . . . . . . . . . . . . . . . . .
5.2 Process description . . . . . . . . . . . . . . . . . . . . . . .
5.2.1 The column streams . . . . . . . . . . . . . . . . . .
5.2.2 The distillation column . . . . . . . . . . . . . . . .
5.2.3 The working principles: thermodynamics . . . . . .
5.3 Binary distillation design . . . . . . . . . . . . . . . . . . .
5.3.1 Problem definition . . . . . . . . . . . . . . . . . . .
5.3.2 External balances: mass balance around the column
5.3.3 Constant Molar Overflow (CMO) assumption . . . .
5.3.4 Operating lines . . . . . . . . . . . . . . . . . . . . .
5.3.5 Number of stages: McCabe Thiele procedure . . . .
5.3.6 Feed quality . . . . . . . . . . . . . . . . . . . . . . .
5.3.7 The reflux ratio . . . . . . . . . . . . . . . . . . . . .
5.3.8 Heat balances . . . . . . . . . . . . . . . . . . . . . .
5.3.9 Final design . . . . . . . . . . . . . . . . . . . . . . .
5.4 Multicomponent distillation design . . . . . . . . . . . . . .
5.4.1 Problem definition . . . . . . . . . . . . . . . . . . .
5.4.2 The external balances issue . . . . . . . . . . . . .
5.4.3 (C 2) assumptions to solve the external balances .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

131
131
131
131
134
135
136
136
139
140
141
141
143
143
146
149
152
154
156
157
157
157
158
159

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

5.5

5.6

5.4.4 Approximate shortcut methods


5.4.5 Fenske equation . . . . . . . . .
5.4.6 Underwood equations . . . . .
5.4.7 Gilliland correlation . . . . . .
5.4.8 Position of feed stage . . . . . .
Non-standard distillation . . . . . . .
5.5.1 Partial condenser . . . . . . . .
5.5.2 Total reboiler . . . . . . . . . .
5.5.3 Open steam (or direct steam) .
5.5.4 Side streams . . . . . . . . . .
5.5.5 Multiple feeds . . . . . . . . . .
Azeotropic column . . . . . . . . . . .
5.6.1 Principles . . . . . . . . . . . .
5.6.2 Extractive distillation . . . . .
5.6.3 Azeotropic distillation . . . . .
5.6.4 Pressure swing distillation . . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

6 Liquid-liquid extraction
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . .
6.1.1 Definitions and aims . . . . . . . . . . . . . . .
6.1.2 Process description . . . . . . . . . . . . . . . .
6.2 Single stage operations . . . . . . . . . . . . . . . . . .
6.2.1 Single stage: problem definition . . . . . . . . .
6.2.2 Single stage: process design . . . . . . . . . . .
6.2.3 Single stage: solvent range . . . . . . . . . . . .
6.3 Multistage operations: cross-current . . . . . . . . . .
6.3.1 Cross-current: cascade configuration . . . . . .
6.3.2 Cross-current: process design . . . . . . . . . .
6.3.3 Cross-current: effect of the number of stages on
ration . . . . . . . . . . . . . . . . . . . . . . .
6.4 Multistage operations: counter-current . . . . . . . . .
6.4.1 Counter-current: column configuration . . . . .
6.4.2 Counter-current: stream definition . . . . . .
6.4.3 Counter-current: process design I . . . . . . . .
6.4.4 Counter-current: solvent rate and position:
and Dmin . . . . . . . . . . . . . . . . . . . . .
6.4.5 Counter-current: process design II . . . . . . .

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
sepa. . . .
. . . .
. . . .
. . . .
. . . .
Smin
. . . .
. . . .

161
162
165
169
170
170
170
172
173
174
177
183
183
184
186
189
191
191
191
191
193
193
194
195
195
195
196
198
200
200
201
202
204
206

This document contains most things found on the hyper-TVT site. It also
contains some errors, and is probably incomplete.

Fundamentals

1.1

Liquid-vapour equilibrium (LVE)

Thermal equilibrium: Heat-transfer stops,


Tliq = Tvap

(1)

Mechanical equilibrium: Forces are balanced, usually that means:


Pliq = Pvap

(2)

Phase equilibrium: rate of condensation = rate of evaporation, no


change in composition. If T and P are constant, equal rates require a
minimum of free energy, and:
li = vi

(3)

If the chemical potentials are the same, the fugacities (because di =


RT d ln(fi ) are also:
fiL (x, T, P ) = fiv (y, T, P )
1.1.1

(4)

General expression for the iso-fugacity condition

In the most general form, the iso-fugacity condition can be expressed as


L
xi i (x, T, P )fi,pure
(T, P ) = yi i (y, T, P )P

(5)

with i = xaii and i = Pfii being the activity and fugacity coefficients, respectively, and a being the activity. i refers to a specific component, and, in the
most general case, the activity and fugacity coefficients depend on temperature, pressure and all compositions (making x and y vector compositions).
Equation (5) can be further simplified for most practical cases, using the
following assumptions:
6

The liquid fugacity of the pure component i can be expressed via its
vapour pressure:
L
fi,pure
(T, P ) = Piv
(6)
The pressure has a negligible effect on the activity coefficient:
i (x, T, P ) = i (x, T )

(7)

The gas is ideal, which means that the fugacity coefficient for the
component i is equal to 1:
i (y, T, P ) = 1

(8)

With these simplifications, the isofugacity condition can be rewritten:


xi i (x, T )Piv (T ) = yi P

(9)

Activity coefficients can be calculated using different correlations, e.g. the


correlations of Margules.

1.2

Binary systems representation

A liquid-vapour equilibrium of a binary system can be graphically represented in different ways:


T vs. x
P vs. x
y vs. x
H vs. x
One of the most convenient is mole fraction in gas phase vs. mole fraction
in liquid phase (y vs. x, usually of the more volatile component). The equilibrium curve can refer to either constant temperature or constant pressure,
with the other value changing along the equilibrium line (which represents
two phases in equilibrium):

In the temperature vs. composition diagram, there are two lines representing
T vs. x and T vs. y: The liquid equilibrium line and the gas equilibrium
line, respectively. There are 3 regions: At high temperatures, only vapour
is present, at low temperatures, only liquid is present. Finally, there is a
region representing a liquid-vapour equilibrium, with different compositions
yi and xi of the vapour and liquid phase.

At T1 , only vapour is present. Moving to T2 in the biphasic region, a certain


amount of vapour condenses. The same kind of diagram is obtained for a P
vs. x representation.
Finally, there is the enthalpy vs. composition diagram.

Isotherms can be drawn between the saturated vapour and saturated liquid
curves, these are called tie-lines. Example:

1.3

Gibbs Phase Rule

The degrees of freedom describe the numbers of variables that can be arbitrarily specified without modifying the system:
degrees of freedom = # variables # constraints

(10)

For non-reacting systems, this relationship can be expressed via the Gibbs
phase rule (with F = degrees of freedom, C = number of components and
P = number of phases):
F =C P +2
(11)
So, for a binary system with 2 phases, there are 2 degrees of freedom. Therefore, 2 variables can be set to uniquely identify the system (e.g. T and P,
T and x, P and x); there are 2 independent variables, while the third is
defined by the other two. The Gibbs phase rule refers to intensive variables
(independent of the size of the system). Examples of extensive variables are
volume, flow rate and number of moles.
1.3.1

Single component systems

Applying the Gibbs phase rule to the following example, we get 2 degrees of
freedom for point A. So, 2 variables can be changed (pressure, temperature)
without changing the phase. The same would happen for point B. For point
C, however, the Gibbs phase rule only returns 1 degree of freedom. Only one
of the variables can be chosen at will, while the other is defined by the first
one. In doing so, one would move along the liquid-vapour equilibrium line
while keeping the same state of the system. Finally, we have the point T:
9

There are no degrees of freedom left, no matter which variable is changed,


the state of the system would change. This point is called the triple point.
The temperature and pressure corresponding to this point are the critical
values, Tc and Pc .

1.3.2

Binary systems

Using the diagram from earlier,

point A would give us 3 degrees of freedom. Since we are at constant pressure, we can still change 2 variables (temperature and composition) and still
have the state of the system remain unchanged. On point B, we would have
2 degrees of freedom, meaning that apart from the already fixed pressure,
we can still change temperature or composition, the remaining variable then
being defined by the former. The same holds for points C and D.

1.4

Raoults law (ideal systems)

Raoults law states, that The partial pressure of a component is equal to


its mole fraction in the liquid multiplied by its vapour pressure. Raoults
10

law is the isofugacity condition for ideal vapour and liquid phases. For a
binary system, it can be expressed as follows:
PA = xA PAv

(12)

PB = xB PBv

(13)

Since it holds, that PA = yA P , PB = yB P and P = PA PB , it follows that:


yA = x A

PAv
P

PBv
P
Raoults law can be shown in a P vs. x diagram:
yB = xB

1.5

(14)
(15)

Distribution coefficients

Raoults law can also be expressed using the distribution coefficients kA and
kB :
Pv
yA
kA =
= A
(16)
xA
P
kB =

Pv
yB
= B
xB
P

(17)

They are strongly dependent on the temperature, since PAv and PBv are also,
according to the Antoine equation (with A, B and C being constants for
pure compounds easily found in the literature):
ln(Piv ) = A

11

B
T (K) C

(18)

1.6

Real systems and Raoults law

In reality, the pressure-composition correlation is not linear, as expressed by


Raoults law, but is much rather represented by curves (green):

To apply Raoults law to real gases, the activity coefficient i can be included:
PA = A xA PAv
(19)
PB = B xB PBv

(20)

where A and B depend on temperature, pressure and composition:


i = i (T, P, xi )

(21)

For i > 1, the deviation from Raoults law is positive and the curves bend
on top, as seen in the previous figure.
For i < 1, the deviation from Raoults law is negative and the curves bend
below, as seen in the next figure.

12

For i = 1, we get an ideal system. At high values of xA , one can use the
ideal version of Raoults law, since the deviation from the real case becomes
very small.

1.7
1.7.1

Constant of relative volatility (CRV)


Binary systems

The constant of relative volatility is the ration of the distribution coefficients


of the two components:
kA
AB =
(22)
kB
which means, that:
AB =

PAv
PBv

(23)

These equations hold for binary, ideal systems following Raoults law. It is
possible to write the following system of equations:
yA = kA xA

(24)

1 yA = kB (1 xA )

(25)

And, combining them:


yA =

AB xA
1 + (AB 1)xA

(26)

The relative volatility is often just a weak function of temperature, and can
therefore be assumed constant in small temperature ranges. Equation (26)
is the equation of a curve and can be drawn in an y vs. x diagram:

13

1.7.2

Multicomponent systems

The most general expression for the distribution coefficient of component i


is:
L
i fi,pure
ki =
(27)
i P
For most practical cases, it simplifies to (using (6) and (8)):
i Piv
P
Or, assuming an ideal mixture when Raoults law is valid:
ki =

(28)

Piv
(29)
P
For multicomponent systems, it is useful to define the constant of relative
volatility with respect to a reference component r:
ki =

i =

ki
kr

(30)

kr =

yr
xr

(31)

with
So we get:
X

yi = k r

i xi

1
kr = P
i i xi
i xi
yi = P
i i xi

1.8
1.8.1

(32)
(33)
(34)

Bubble and dew point


Bubble point

The bubble point is the temperature, at which a liquid mixture starts to boil.
At a defined pressure and composition, bubble and dew point are fixed. The
bubble point TBP fulfils the following equations:
X
yi = 1
(35)
i

yi = ki (TBP )xi

(36)

It can be calculated iteratively. In order to do this, one must first estimate a


value for TBP . From this, ki (TBP , P, xi ) is calculated (using the known total
pressure and Antoine equation constant). Finally, the iteration continues
until:
X
k i xi = 1
(37)
i

14

1.8.2

Dew point

The condensation temperature TDP fulfils the following system of equations:


X
xi = 1
(38)
i

xi =

yi
ki (TDP )

(39)

Like the bubble point, it can be determined iteratively, the one difference
being the exit condition:
X yi
=1
(40)
ki
i

1.9

Henrys law

As we saw earlier in the pressure-composition diagram, we can use Raoults


law for high compositions, since the linear behaviour coincides with the
tangent to the curve. For xi << 1, Raoults law is not valid, and we apply Henrys law: The partial pressure of component i in the gas phase
is proportional to its mole fraction in the liquid phase. The constant of
proportionality is the Henrys constant (H).
Pi = Hxi

(41)

Henrys constant can be seen as the vapour pressure of the pure compound at
infinite dilution. As can be seen below, Henrys law is a good approximation
at very low concentrations of A:

We can say, that


yi =

Hi
xi
P

15

(42)

Hi
= mi
(43)
P
where mi is the slope of the equilibrium line in the y vs. x - diagram:

The larger the slope, the more willing a component is to leave the liquid
phase for the gas phase. This corresponds to the temperature dependence
of Hi :
E
Hi (T ) = H0 exp(
)
(44)
RT
So, Hi increases with temperature, as does the slope. Absorption is therefore
favoured at low temperatures and high pressures.

1.10

Ternary systems representation

Ternary systems are usually represented using either right triangular or equilateral triangular diagrams. Any point in the following example represents
a different composition of the system:

16

In particular:
The points at the three apexes A, B and C represent the three pure
compounds.
Any point on a side of the triangle represents a binary mixture of the
two points that enclose the side.
Any point in the triangle (e.g. S, R) represents a mixture of the three
components A, B and C.
The sum of the lengths of the perpendicular lines drawn from any point
in the diagram is equal to the height of the triangle. So, if the height is
scaled from 0 to 100, the length of the line perpendicular to (for example)
the side opposite of A corresponds to the percentage of A in the mixture.
The contents of B and C can be determined in the same way.
The degrees of freedom can be determined using the Gibbs phase rule. At
the point S (outside the immiscibility region), we have 4 degrees of freedom.
Since pressure and temperature are set, this leaves us with 2 compositions
we can chose. The third one is then determined. A point in the immiscibility
region, like R, only has 1 degree of freedom left (after subtracting pressure
and temperature). The compositions of the two phases are determined by
the end points of the tie line (M, P).
The curve LMNOPQ represents the saturation curve for the system, also
called solubility envelope. Herein, the three components are not fully miscible, and any system inside this immiscibility region (like R) will split into
17

two phases in equilibrium to both ends of the corresponding tie line. Point
O is called the plait point, the two phases have the same composition and
become one (the tie line becomes a point). The solubility envelope changes
with different temperatures and pressures. For example, it becomes smaller
for higher temperatures:

When the process design of a ternary system uses graphical methods, it is


often more convenient to employ a right triangular diagram, e.g.:

A diagram typically doesnt contain all tie lines that are needed for the
graphical solution of problems. Additional tie lines can be constructed given
an auxiliary line.

18

One starts with the given blue auxiliary line, B-K. Then, choosing any point
P on this line, parallels through P to A-C and B-C are drawn. The resulting
intersections of these two lines with the edge of the immiscibility region gives
one tie line. Reversely, given a couple (the more lines, the more precise) tielines, one can reconstruct the auxiliary line. The same procedure can be
followed with right right triangular diagrams.

1.11

Azeotropic systems

Sufficiently non-ideal systems with relatively close boiling points often show
azeotropic behaviour (B and C):

Not all non-ideal systems have to show azeotropic behaviour. For example,
systems can have only positive ( > 1)

19

or negative ( < 1) deviations from ideality.

When substances have a very large positive deviation from ideality (very
large ), the liquid mixture has a minimum boiling point at the azeotropic
composition:
The liquid mixture boils, but the composition of the liquid and gas phases
do not change any further

20

The vapour and liquid compositions are identical, and the distribution coefficients and k-values of all components are equal to 1.

The three previous and next figures represent homogeneous azeotropes, at


the azeotropic composition L, there is only one liquid phase.
Negative deviation from ideality leads to maximum boiling point azeotropes:

21

Heterogeneous azeotropes exists when the two components are only partially miscible as liquid. A vapour mixture at the azeotropic composition
L, condenses into two different liquid phases at equilibrium compositions
respectively K and M.

Heterogeneous azeotropes are always minimum boiling, because of the very


large positive deviation from ideality which is necessary to cause the separation of the liquid phases. All presented systems are binary, so no more
than one heterogeneous azeotrope can exist (one vapour phase can coexist
with a maximum of two liquid phases).

1.12

Fundamental principles of mass transfer

Mass transfer is described as the movement of a component from one location to another due to disequilibrium, between phases or in the same phase.
There are two basic mechanisms for mass transfer, molecular diffusion and
turbulent diffusion, where molecular diffusion is a lot slower than its turbulent counterpart. The first one happens due to spontaneous microscopic
motion of individual molecules, while the second one is a random, macroscopic fluid motion. Here, only molecular diffusion due to concentration gradients will be considered, focussing on binary systems. Other driving forces
22

other than the aforementioned concentration gradient will not be considered. Most systems treated in this course have mass transfer taking place
across a two-phase interface. Therefore, it is important to choose the equipment in such a way, that the interface between the two exchanging phases
is maximised.

1.13

Ficks law

Assuming a system of homogeneous fluid composition is divided by a wall.


Adding particles to one of the two divisions, then removing the wall, shows a
concentration gradient (after some time). A disequilibrium had been created
within the same phase, and the particles started moving in direction of lower
particle concentration z.

This phenomenon, molecular diffusion, stops, when the red particles are
uniformly distributed in the phase. The flux of components can be described
by Ficks law:
i
dci h
Ni = Di
(45)
mol s1 cm2
dz
The orders of magnitude for the diffusion coefficient:
Liquids: Di = 105 cm2 s1
Gases: Di = 101 cm2 s1
Solids: Di = 1010 1013 cm2 s1

1.14
1.14.1

Film model
Steady state ordinary molecular diffusion

In separation processes, it is very important to consider the mass transfer


across an interface between a gas and a liquid (absorption, stripping, flash
evaporation, distillation) or between two liquids (liquid liquid extraction).
The key concept of the film model is that all resistances to mass transfer
between two phases reside in the phases themselves, there is no resistance
to crossing the actual interface. The resistances occur in small regions to
both sides of the interface.
23

The film model can be viewed as a membrane separating two bulk phases of
uniform compositions. Concentration differences exist between the sides of
the membrane and act as the driving force of mass transfer. The membrane
represents the liquid film, and not the interface!
1.14.2

One-film model

In the one-film model, only the film on the liquid side is considered for
the mass transfer resistance. We assume to have a pure gas of components
A diffusing (absorbing) into a nonvolatile liquid B. Therefore, there is no
resistance on the gas side and at the interface, the equilibrium condition is
given by Henrys law:
cA,i = HA PA
(46)
The assumption is made that all molecules of A diffusing into the liquid film
will diffuse into the bulk of B.

Due to the film being very thin, this assumption is valid. Bulk flow of A is
neglected, so all transfer of A is due to molecular diffusion.
Applying the assumption of a steady-state to Ficks law (45), we get:
A
d(DAB ( dc
dNA
dz ))
=
=0
dz
dz

24

(47)

d2 cA
=0
dz 2
or, using the mole fractions xA = ccA
NA = DAB

dxA
dz

(48)

(49)

and applying the material balance


dNA
= 0 with xz=0 = xAi and xz= = xA,bulk
dz
we get:
NA = DAB c

xA,i xA,bulk

(50)

(51)

The ratio DAB


can be replaced by the mass transfer coefficient kc to give:

NA = kc c (xA,i xA,bulk )

(52)

NA = kc (cA,i cA,bulk )

(53)

Therefore, the one-film model postulates a linear concentration gradient


within the film. The rate of mass transfer, NA is directly proportional to the
concentration difference, which acts as the driving force for mass transfer.
1.14.3

Two-film model

The two-film model assumes that resistances occur in both a small film on
the liquid side of the interface as well as in a small film on the gas side of the
interface. Again, the interface adds nothing to the resistances, and we can
assume that the liquid and gas phases are at equilibrium at the interface
yi = feq. (xi )

(54)

and, if Henrys law can be applied, that the equilibrium can be assumed
linear:
yi = mxi
(55)
At steady state, with no accumulation possible, we can say that
N = kG cG (y yi ) = ky (y yi )

(56)

N = kL cL (xi x) = kx (xi x)

(57)

for the gas and liquid side, respectively, where ky and kx are the locally
applicable mass transfer coefficients. Since equations (56) and (57) must be
equal, it is possible to write:
ky (y yi ) = kx (xi x)
25

(58)

The ratio

(y yi )
kx
=
ky
(xi x)

(59)

is the slope connecting the operation line to the equilibrium line, or, in other
words, the line connecting equilibrium composition with bulk composition.

Therefore, if we know the local mass transfer coefficients, we could calculate


the rate of mass transfer. Unfortunately, it is impossible to get close enough
to the interface to experimentally determine these coefficients. Therefore, a
different expression for N has to be found.
1.14.4

The overall mass transfer coefficient (MTC)

In order to get rid of the interface concentrations, we express the driving


force via. equilibrium concentrations using Henrys law:
y = y y = y mx

(60)

N = Ky (y y )

(61)

With Ky being the overall mass transfer coefficient. It can be related to the
local mass transfer coefficients via the following steps:
N = Ky ((y yi ) + (yi y ))

(62)

N = Ky ((y yi ) + m(xi x))

(63)

It still holds, that:


y yi =

N
ky

(64)

xi x =

N
kx

(65)

and, finally:
1
1
m
=
+
Ky
ky
kx
26

(66)

1
1
1
=
+
Kx
mky
kx

(67)

If the gas diffusing into the liquid is very soluble, absorption is favoured,
the slope of the equilibrium line, m, is very small, and the second term in
equation (66) also becomes very small. It then holds, that:
1
1
=
Ky
ky

(68)

For the opposite case, if the gas is not soluble, then the slope of m becomes
very large. The first term in equation (67) can be neglected and we get:
1
1
=
Kx
kx

1.15

(69)

Numerical methods - principals

It is often required to find the root(s) for a problem of the form


f (x) = 0

(70)

since all analytical expressions can be written in this form. Following are
two numerical methods for solving problems of this kind.

1.16

Half interval method

This method is generally used when there is no analytical expression for


the derivative of f (x) with respect to x available. The algorithm works as
follows:

27

Of course, choosing a and b for each subsequent iteration uses xnew as either
a or b. Also, a requirement for a and b is:
f (a)f (b) < 0

1.17

(71)

Newton method

This algorithm is applied if an analytical expression for the derivative of the


function is available, it is significantly faster than the half interval method.
First, a starting value xi is chosen (x0 in the graph below). The derivative
f 0 (x) is then calculated. The condition is, that
y = f (xi ) + f 0 (xi )(x xi ) = 0

(72)

from which a value for the next iteration can be found:


xi+1 = xi

f (xi )
f 0 (xi )

This equation can also be obtained via the equations for the tangent.

28

(73)

1.18

Lever-arm rule

The lever-arm rule is based on the same principle as the see-saw momentarm balance. In equilibrium, it has to hold that:
(weight F1 ) distance(F1 M ) = (weight F2 ) distance(M F2 )

(74)

The lever-arm rule can be applied to mixing calculation on enthalpy vs.


composition as well as on ternary diagrams for extraction. In the second
case, the lever-arm rule can be used to easily solve overall mass balances for
two mixing streams.

Doing the overall mass balance, it becomes apparent that F1 , F2 and M lie
on the same straight line.

29

The mass of the two streams F1 and F2 is then proportional to their respective distance to the point M from the point representing the other stream
(shorter distance M F2 means heavier F1 ).

Contactors

2.1
2.1.1

Gas-liquid contactors
Principles

Gas-liquid mass transfer takes place across an interface at the phase boundary over a certain contact time. Any equipment used for this kind of operation should provide a high interface area. Additionally, proper mixing is
also a requirement to be considered in the design of a good contactor.
2.1.2

Equipment overview

Many types of column designs are available: Tray columns, packed columns,
spray towers, bubble columns, centrifugal contactors, ... . The mainly differ
in the choice of continuous phase, which can be either the gas or the liquid,
and in the mode of flow, which can be co-current, counter-current or crosscurrent. In the end, they all must enable a sufficient rate of mass transfer
by providing adequate mixing and contact time.
Tray columns are cylindrical vessels, which operate in either cross- or
counter-current operation and can be operated under pressure. The mass
transfer takes place in separate trays.

30

Packed columns are vertical cylindrical vessels containing packing material between sections 1 and 2 in the following figure. The liquid flows down
the column via gravity, while the gas rises from the bottom. Mass transfer
takes place between the liquid film coating the packing area (thereby greatly
increasing the contact area) and the gas bubbling through from the bottom.
Packed columns can be operated under pressure.
Spray columns are cylindrical vessels filled with gas through which the
liquid is sprayed. They can be advantageous when the solute is very soluble
in the liquid phase.

Bubble columns are cylindrical vessels filled with liquid through which
the gas phase is bubbled.
Tray and packed columns are the devices most used in separation technology
using gas-liquid mixtures.

2.2
2.2.1

Tray columns
Description

Tray columns can be operated under pressure and consist of a cylindrical


vessel in which gas and liquid flow either flow in a cross-current or countercurrent fashion to each other. Mass transfer occurs on a series of metal trays
or plates.

31

In counter-current operation, gas and liquid flow through the same perforations, the plates take up the whole cross-section of the column. In crosscurrent operation, the liquid flows over the plate and down a downcomer,
and only part of the cross-section is taken up by the plate. The operating
range and liquid flow pattern may be controlled for a better mass transfer
efficiency via the area of the downcomer, which leads to cross-current systems being more widely used. The number of downcomers is not limited to
one. In a double pass plate, for instance, two downcomers are arranged, so
that the horizontal direction of liquid flow changes with every plate.

On the other hand, the more downcomers are employed, the lower the area
available for gas-dispensers (perforations in the plate) becomes.
Due to plate disposition and stream flow pattern, different zones are present
around a plate, with different regimes.

32

In the above figure, zone A is the downcomer from the above plate. Liquid
enters the tray here as a continuous phase and flows through the zone aerated
by the upflowing gas stream (tray), before departing to the next tray on
the right hand side (D). Zone B is a continuous liquid directly above the
perforations of the plate. There is a froth (bubbly, or aerated, mixture of
vapour and liquid, with observable height) over the tray. Liquid droplets
are ejected into the gas phase above the froth. Under specific circumstances
(high volumetric vapour to liquid flow), this zone can invert to a continuous
vapour spray, meaning that the vapour continuous phase is extended and its
liquid content will decrease with height (distance to the tray). A properly
designed and functioning sieve tray operates in froth contacting mode if at
all possible. Zone C describes the continuous vapour phase containing liquid
droplets. The next downcomer is shown as D, with the only difference to A
being the location of entry (right vs. left). Finally, zone E describes the gas
flow to the plate.
2.2.2

Tray types

The most common and used plates are:


Sieve plate

33

A sieve plate is usually a plate with simple round perforations. Upflowing gas prevents the liquid from flowing through these perforations, although with low gas rates, this can still occur (weeping phenomenon). Different kinds of perforations can be used to increase
performance.
Valve plate

The valve plate-design minimises the phenomenon of weeping. It consists of a cap covering the perforation and legs to limit the vertical
rise of said cap. The lower the gas flow, the lower the rise of the valve,
although this also depends on pressure. This allows for control of the
amount of liquid draining to the plate below and therefore also of the
liquid retention time on the plate.
Bubble- (or bell-) cap plate

34

A bubble-cap plate consists of fixed caps mounted over the perforations. The caps have rectangular or triangular slots cut around its
side. The gas flows up through the perforations, rises, turns around
and passes out through the slots of the cup directly into the liquid,
producing a froth regime. The caps are well sealed to the plate
surface, preventing the liquid from draining at low gas flow rate.
Tray comparison Plates types can be compared on the basis of
cost, pressure drop, efficiency, vapour capacity and flexibility.

Usually, sieve plates are preferred because of their low cost, unless
flexibility is required. Because of the high associated pressure drop,
bubble caps are rarely used, although they have been widely used in
the past, especially for cross-flow plates. They have now largely been
replaced by valve plates.
2.2.3

Operative conditions

Four regimes can be distinguished in a tray column, depending on flow rates


and continuous phase:
1. Bubbles regime

35

The liquid is quiescent and the gas bubbles through it. This is the
case at low gas flow rates. Due to the poor mixing of gas and liquid,
this case exhibits low stage efficiency. This low efficiency, coupled
with the low gas flow rate, makes the bubbles regime undesirable for
commercial applications.
2. Foam regime
At higher gas flow rates compared to the bubbles regime, the gas rises
faster and a foam is formed on top of the liquid. The stage efficiency is
increased by a higher interfacial area (more bubbles), though, should
the foam become too much and too stable, the phenomenon of entrainment becomes excessive: Liquid is carried to the tray above, the
stage efficiency drops and the column may even become flooded (filled
with liquid) and inoperative. An additional chemical antifoam agent
is therefore required.
3. Froth regime
The liquid is the continuous phase and the gas passes through it as
jets or bubbles, with the surface of the liquid boiling and splashing. A
very high mass transfer due to a very good liquid-gas contact makes
this the most common and favoured regime.
4. Spray regime
At very high gas flow rates, the gas is the continuous phase and the
liquid is sprayed into it as fine droplets. Because of the poor liquid
mixing, mass transfer rates are usually low in this regime. A shift
from froth to spray regime can result in a significant reduction in
stage efficiency (e.g. from 65% to 40%).

2.3
2.3.1

Packed columns
Description

Packed columns are cylindrical vessels containing one or more sections of


packing material. The liquid moves downward through gravity, while the
gas flows upward in a counter-current fashion. The packing material aims to
36

maximise the gas-liquid interfacial area to enhance mass transfer. Packing


can be either structured (a.) or random (d.). The packing material rests
between a lower support plate and an upper hold-down plate (b.), which
aim at avoiding packing movement. The liquid distributer (c.) is situated
above the hold-down plate, and ensures even distribution of the liquid flow.

2.3.2

Packing types

The types of packing material can be split into random packing and structured (aka. ordered, stacked or arranged) packing. A good packing material
should have a high surface area to volume ratio, since mass transfer takes
place between the liquid layer coating the packing material and the gas
flowing through the packing elements. Additionally, a high liquid flow capacity through the packing layer is required to uniformly coat the packing
material with liquid (good wettability) as well as good resistance to pressure (high strength) as well as a low pressure drop. The larger the size of
the packing material units, the lower the pressure drop, but also the mass
transfer. Optimal packing materials compromise between the two. Metal
and plastic packing material has largely replaced the ceramic of the old days.
For random packing, the old rashig rings (d3) and berl saddles (d2, yellow pieces) are substituted by pall rings, intalox saddles, mini cascade ring,
and a broad range of other different shaped elements (d1, d2). However,
ceramic elements are still used when dealing with corrosive materials and
good wettability is required. Metal elements provide good wettability and
37

high strength. Plastic elements, on the other hand (usually polypropylene)


have poor wettability and fail at high temperatures. Due to their low cost,
they are nevertheless widely used.
Structured packing is a lot more cost intensive than random packing, but
has the advantage of higher mass transfer as well as a very low pressure
drop.

2.3.3

Comparison between packed columns and tray columns

Packed columns offer a lower operating range:


Too low liquid flow rate: Wetting may not be sufficient.
Too hight liquid flow rate: Plate columns are often more economic.
If solid particles are present in the liquid: Cleaning of the packing is
very expensive and complicated compared to plate columns.
Working under stressed temperature and pressure conditions: Packing
elements can easily break.
On the other hand, packing can also be the more favourable option:
Costs: Packed columns are usually less expensive for small column
diamenters (< 0.6 m).
Handling of corrosive chemicals: Cheap ceramic or other chemically
resistant packing can easily be used.
Foaming liquid: Foaming liquid can be handled more easily in packed
columns because of the lower agitation of the liquid by the gas.
38

2.4

Liquid-liquid contactors

Liquid-liquid contactors employ agitation of the liquid in order to enhance


mass transfer by increasing the dispersion of one phase in the other and
increasing the interfacial area. The agitators create mixing zones, which
alternate with settling zones along the length of the column.

Various commercially available column designs for liquid-liquid contactors


mainly differ in the design of the agitators and stator disks. (Respectively
a and b in the above figure). The goal of the extraction process is to purify
an aqueous stream (heavy, dispersed phase) by extracting a solute with a
water-immiscible organic solvent (light, continuous phase). The pollutant,
having a higher affinity for the organic phase, is transferred to the latter
along the length of the column.

Flash distillation

3.1
3.1.1

Introduction
Definitions and aims

Flash evaporation is one of the simplest unit operations. A liquid stream is


partially vaporised in a flash drum at a certain pressure and temperature.
The result is vapour, which is richer in the more volatile component than
39

the remaining liquid. The opposite operation to flash distillation (partial


evaporation) is partial condensation (with a complete vapour feed). Knowledge of the dew point of the mixture is very important.
Usually, flash distillation cannot achieve a large degree of separation, and is
therefore employed as an auxiliary operation to prepare streams for further
processing. In some cases however, like the desalination of sea water, complete separation can be achieved.
If only two components are present in the feed, the flash is called binary,
while more than two components in the feed define a multicomponent flash.
3.1.2

Process description

The flash can be seen as a distillation with only one equilibrium stage. The
operation stops, when the liquid and vapour streams reach the equilibrium
compositions defined by temperature and pressure, and the two streams can
easily be separated.

The incoming liquid is first heated and pressurised, before being fed into the
drum. Due to the large pressure drop, the liquid evaporates very quickly
(hence flash).

3.2
3.2.1

Binary flash
Problem description

In the case of a binary flash, the feed consists only of two components: A
and B. One of the key assumptions here is, that the flash drum acts as a
single equilibrium stage, so liquid and vapour are in equilibrium at the end

40

of the flash operation.


The goal is now either to design a new flash unit for a new separation or to
verify that an existing flash drum is unit is good to carry out an existing
one.
3.2.2

Data, specifications, unknowns

Using the Gibbs phase rule, we receive 2 degrees of freedom for the system,
which should translate to 7 equations and 5 variables. Let us check that for
the dashed line in the figure.

Usually, the design of a flash drum consists of determining the temperature


and pressure at which the flash takes place, the amount of heat provided
and finally the liquid and vapour stream compositions and flow rates. This
gives us:
Unknowns: Tdrum , Pdrum , QH , V, y, L, x
For the sake of simplicity, x and y refer to the more volatile component.
The data of the feed mixture is usually known:
Data: TF , PF , F, zF
For this system, the following equations can bewritten:
Phase equilibrium equations (using eq. (12), Raoult):
yP = xPAv (T )
(1 y)P = (1
41

x)PBv (T )

(75)
(76)

These equations can also be written using the distribution coefficients (via
eq. (29):
y = xkA (T )

(77)

(1 y) = (1 x)kB (T )

(78)

Material balance equations:


F =L+V
F z = Lx + V y

(79)
(80)

Energy balance equations:


F hF + QH = LhL + V HV

(81)

So, 5 equations and 7 variables confirms our 2 degrees of freedom.


In order to solve our system of equations, the two degrees of freedom must
be specified, meaning that 2 of the 7 unknowns must be fixed. This choice
depends on the availability of the data. In the aforementioned cases of process design and process verification, the availability of starting data differs.
Additionally, in real life industrial situations there are often limitations on
operating conditions, and therefore temperature and pressure are often not
a free choice. Usually, one of the two specifications in a flash drum is the
pressure. For the last degree of freedom, several approaches can be considered, depending on process design or process verification.
It is important to note, that the equilibrium and mass balances contain
variables, which do not appear in the energy balance equation and the other
way around. In this case, the two sets of equations are decoupled. In the
sequential procedure of solving the system, we will go through the first group
of equations (equilibrium and mass balance) and solve the system for L, V ,
x, y and after this, the energy balance would be solved to get QH .
3.2.3

Process design

In this case, the given task is to design a new flash unit for a required
separation. Given the following system data:
Unknowns: Tdrum , QH , V, y, L, x
Data: TF , Pdrum , PF , F, zF

42

We assume that the pressure of the flash drum is known. The system now
has one degree of freedom left, and we need to specify this variable to unequivocally determine the system. The following possibilities, which variable
to specify, can be considered:
a. Vapour mole fraction y
b. Liquid mole fraction x
c. V /F : Fraction of vaporised feed, also indicated as .
d. L/F : Fraction of feed remaining liquid.
e. Temperature in the drum, Tdrum
1st case: x or y is known. The solving procedure is an iterative method
until convergence:
1. Assume a temperature Tdrum and calculate y (or x) from one of the
equilibrium equations.
2. Verify, if the found composition value also fulfils the second equilibrium
equation.
3. Find L and V from the mass balance equations.
4. Solve the energy balance equation for QH .
2nd case: V /F or L/F is known.
balance equations

With this, and the overall mass

F =L+V
F z = Lx + V y

(79)
(80)

it is possible to draw the operating line


y=

L
F
x+ z
V
V

43

(82)

and find x and y graphically.


The temperature can now be found from the ratio:
P v (T )
y
= A
(83)
x
P
The energy balance is an independent equation, with the unknown QH . hL
and HV can be found from an enthalpy-composition diagram of the binary
mixture,

while hf must be calculated from the following equation (since the feed is
neither a saturated liquid, nor has equilibrium composition):
hf = xA Cp,A (T Tref ) + xB Cp,B (T Tref )

(84)

3rd case: Tdrum is known. If Tdrum is known, PAV (T ) and PBV (T ) can
be calculated. Using P , we can now calculate x and y using the phase
equilibrium equations
yP = xPAv (T )
(1 y)P = (1
44

x)PBv (T )

(75)
(76)

and, subsequently, L and V can be calculated via the mass balance equations. QH can be calculated from the independent energy balance.
Very often, the distribution coefficients kA (T ) and/or kB (T ) are given. They
are related to the vapour pressures via equations (16) and (17):

3.2.4

kA =

Pv
yA
= A
xA
P

(16)

kB =

Pv
yB
= B
xB
P

(17)

Process verification

The purpose of the verification problem is to check, how good a separation


can be achieved in an existing flash unit. The following system is given:
Data: TF , PF , F, zF , Pdrum , QH
In this case, QH is a known value. This is often true, when the flash drum
is insulated and process is assumed to be adiabatic, i.e. QH = 0.
Unknowns: Tdrum , V, y, L, x
The solving procedure in this case follows an iterative path:
1. Assume a temperature Tdrum and calculate x and y from the equilibrium equations (75) and (76).
2. Calculate L and V from the mass balance equations (79) and (80).
3. Verify, if the found values fulfil the energy balance equation (81).

3.3
3.3.1

Multicomponent flash
Problem description

If the feeding stream has more than 2 components, an analytical procedure is


needed. The key assumption, that the flash drum acts as a single equilibrium
stage, implies that the vapour and liquid streams leaving the drum have
equilibrium composition.
3.3.2

Data, specifications, unknowns

The system scheme looks like the one for the binary system:

45

For this system, we can write:


Unknowns: Tdrum , Pdrum , QH , V, yi , L, xi = 2C + 5 in total, where
i = 1, 2, ..., C are the components.
The data of the feed is known:
Data: TF , PF , F, zi
and the following equations can be written:
Phase equilibrium equations: C independent equations.
yi = ki xi

with i = 1, ..., C

(85)

The phase equilibrium equations are usually written in terms of distribution coefficients:
ki = ki (Tdrum , Pdrum , xi )

(86)

For an ideal system, ki does not depend on the mole fractions and the
above equation reduces to:
ki = ki (Tdrum , Pdrum )

(87)

Stoichiometric rules: 2 independent equations.


C
X
i=1
C
X
i=1

46

yi = 1

(88)

xi = 1

(89)

Material balance equations: C independent equations.


F =L+V

(90)
with i = 1, ..., C 1

F zi = Lxi + V yi

(91)

Energy balance equations: 1 independent equation.


F hF + QH = LhL + V HV

(92)

Combining equations (88), (89), (110), (91) and (92), we get a total of
(2C + 3) independent equations. This confirms, that a multicomponent
flash (like a binary flash) has 2 degrees of freedom [(2C + 5) unknowns (2C + 3) equations].
As discussed for the binary case, one of these two degrees of freedom is
taken by the pressure of the flash drum, Pdrum , while for the last remaining degree of freedom, the choice depends on the available data, which in
turn depends on the kind of problem (design vs. verification).
The following procedure can be used to solve the system of equations:
First, use the phase equilibrium equations (85)
yi = ki xi

(85)

and replace yi in the mass balance equation (91):


F zi = Lxi + V ki xi

(93)

solving for xi :

F zi
L + V ki
and substituting L by the mass balance equation (110):
xi =

xi =

F zi
F V + V ki

(94)

(95)

It is common to divide the numerator and denominator by the feed rate F,


we get expressions for xi and yi :
xi =

zi
1 + (ki 1) VF

yi =

ki zi
1 + (ki 1) VF

(96)

The ratio VF is usually denoted by the Greek letter , because it is bounded


between 0 and 1. Applying the stoichiometric relations (88) and (89)
C
X

C
X

xi = 1

i=1

i=1

47

yi = 1

(97)

equations (96) can be rewritten as:


C
X
i=1

C
X

zi
=1
1 + (ki 1)

i=1

ki zi
=1
1 + (ki 1)

(98)

and, subtracting equations (98) from each other:


C
X
i=1

X
ki zi
zi

=11
1 + (ki 1)
1 + (ki 1)

(99)

i=1

C
X
i=1

(ki 1)zi
=0
1 + (ki 1)

(100)

In order to solve equation (100), two cases can be considered:


ki are known: Find the root to the function f () = 0 to find .
Once is known, equations (96) are used to calculate xi and yi .
is given: Equation (100) is used to find Tdrum , following an iterative
procedure:

.
1. Assume an initial value Tdrum

2. Calculate ki with the assumed Tdrum


.

3. Calculate the root of the function f (Tdrum ) = 0.

4. If the root coincides with the assumed Tdrum


, this is the value for
Tdrum . If not, start again at point 1.

For the analytical search of the root of a function f (x) = 0, one of the
numerical methods, presented in the basics section, can be used.
3.3.3

Split factor

If the feeding stream contains more than 2 components, an analytical procedure is needed.
The split factor is defined as:
Si =

V yi
Lxi

(101)

Using the well-known equilibrium equation


xi Piv = yi Pi

48

(102)

this can be rewritten as


Si =

V Piv
LPi

(103)

With the overall mass balance, we get:


F zi = Lxi + V yi

(104)

Dividing by Lxi , the split factor appears:


V yi
F zi
=1+
= 1 + Si
Lxi
Lxi

(105)

This can be rearranged to


Lxi =

F zi
1 + Si

F zi Si
1 + Si

(106)

Lxi = L

(107)

V yi =

from which we can obtain:


C
X

C
X

V yi = V

i=1

3.3.4

i=1

Bubble point

It is sometimes important to know the bubble point of the liquid feed. The
flash drum is used to bring the liquid to its boiling point (the very first drop
becomes vapour). At this point, no vapour is produced yet and = VF = 0.
Of the two degrees of freedom that we know this system has, one is the
pressure (as usual), while the other one will be defined by . zi now coincides
with xi , since everything remains in the liquid phase. Once we know xi , we
can make an assumption for TBP and follow an iterative procedure.
3.3.5

Dew point

Analogous to the bubble point. In this case, the feed is entirely in the vapour
phase and the purpose is to bring it to the dew point. No liquid is produced,
and = 1, and therefore also zi = yi . Knowing xi , we again assume TBP
and proceed iteratively.

3.4
3.4.1

Nonideal flash
Process design

In case of a real system, the previous assumed independence of the distribution coefficients from the mole fractions is no longer valid:
ki = ki (Tdrum , Pdrum , xi )
49

(86)

Using the isofugacity condition


fiL (x, T, P ) = fiv (y, T, P )
liquid fugacity = vapour fugacity

(4)
(108)

and assuming an ideal gas, we can write:


xi i (P, T, x)Piv (T ) = yi P

(9)

with i being the activity coefficient which can be obtained from the Margules correlations.
So, for a multicomponent, ideal flash, we get:
ki (P, T, xi ) =

P v (T )
yi
= i (P, T, x) i
xi
P

(109)

We can write the following equations:


Material balance equations: C independent equations.
F =L+V

(110)
with i = 1, ..., C 1

F zi = Lxi + V yi

(111)

Phase equilibrium equations: C independent equations.


ki =

yi
xi

with i = 1, ..., C

(112)

Stoichiometric rules: 2 independent equations.


C
X

yi = 1

(113)

xi = 1

(114)

i=1
C
X
i=1

As with the ideal case, everything can be rewritten in the form of


equation (100):
C
X
i=1

(ki 1)zi
=0
1 + (ki 1)

(100)

Again, this system is solved iteratively:


1. Assume an initial value for the mole fraction xn . Usually, one
assumes an ideal system for this.
50

2.
3.
4.
5.

3.5
3.5.1

From this mole fraction, calculate ki (P, T, xn ).


With this, use equation (100) to find .
From the mass and equilibrium balances, recalculate xn .
If xn is close enough to xn , it is the true value. If not, the iteration
is repeated from step 1 until convergence.1

Flash cascades
Flash operations at constant temperature

As noted previously, flash operations can be seen as distillations with only


one equilibrium stage. For this reason, high degrees of separation usually
cannot be achieved. Therefore, a flash is usually the first process in industrial processes in order to achieve a crude separation before feeding the
stream to a distillation or other separation/purification process.
Due to the parallels between flash and distillation, it is possible to describe
the latter through a series of units of the former.

In this figure, all units operate at the same temperature but different pressure. Towards the top, the stream becomes richer in the most volatile component until the purified stream V1 is obtained. On the bottom, the opposite
51

happens, and the least volatile component is collected in L5 . Additionally,


other undesired streams (Ln6=5 , Vn6=1 ) are produced and can be recycled (after repressurising) to the previous flash stage in order to reduce waste.
For a binary system, this flash cascade can be described in a P vs. x diagram.
The composition of the liquid (or vapour) stream fed to the cascade of stages
(green points in the figure below) is very close to the composition of the
recycled streams to the same stage. Obviously, the total feed composition
(not shown) will lie somewhere between these two, depending on the flow
rates of the two streams.

The flash cascade can be a useful operation. However it becomes unattractive when several stages are needed to reach the specifications on operating
and economical considerations. Every stream of the flash cascade must be
pressurised to reach the conditions for the following flash operation, and not
needed outlets must be re-pressurised back when recycled to the previous
stage.
3.5.2

Flash operations at constant pressure

The other option that presents itself would be to keep the pressure constant
while changing the temperature, shown in the following T vs. x diagram.

52

Again, the composition of the liquid or vapour stream fed to the cascade is
indicated by the green points, and the total feed composition (not shown)
will lie somewhere between the compositions of recycle and feed, depending
on the flow rates. In this case, feeds leaving the individual flash units have
to be heated up (in the case of liquid streams ) or cooled down (in the case
of vapour streams).

53

Again, the flash cascade allows for a better degree of separation, while on
the other hand the constant re-heating can make it very inconvenient.

Absorption and stripping

4.1
4.1.1

Introduction
Definitions and aims

Absorption is a unit operation in which one or more selected components,


the solutes, are removed from a gas mixture (solute + carrier gas) by contact
with a nonvolatile liquid stream, the solvent. The carrier gas is assumed to
be insoluble in the solvent, so only the solute moves to the liquid phase.

Mass transfer of solute to the solvent is due to its greater affinity towards
the liquid phase compared with the gas phase. The transport of the solute
requires that the two phases are in contact for a certain contact-time.
The affinity can be of physical or chemical nature. In the latter case a reaction is present between solute and solvent termed as chemical absorption.
An example of this is the removal of CO2 or H2 S removal by NaOH through
reversible reaction. An irreversible reaction occurs when using MEA (monoethanolamine) instead of NaOH. In the case of irreversible reactions, the
resulting solvent must be disposed of, whereas in reversible reactions, the
solvent can be regenerated. Therefore, reversible reactions are often preferred.
In physical absorption only physical interactions are present between solute
and solvent. An example is the transport of acetone from an air stream to
a water stream. The acetone dissolves into the water while the air passes out.
Stripping is the opposite operation of absorption. In this unit operation,
one or more components of a liquid mixture are removed by evaporation
into an insoluble gas stream. The same assumptions as in the absorption
case are valid here. i.e. the carrier liquid is nonvolatile and the gas is insoluble into the solvent. Only the solute moves from the liquid to the gas stream.
54

The aim of these operations is the separation of the inlet stream (gas for
absorption and liquid for stripping) from one or more components mixed in
it. This may be needed for one of the following reasons:
These components (the solutes) are pollutants and they have to be
removed from the gas or liquid stream before it is released to the
environment.
They are valuable components and they have to be recuperated before
the stream is wasted.
They are intermediate products and they must be separated from the
stream to be used in further productions.
However it is interesting to note that in these operation units, while the
required stream is being purified, another one is being polluted with the
separated components, which moved into it. For this reason, as we see in
the next section, stripping and absorption are two very linked processes.
4.1.2

Process description

The most typical configuration for an absorption/stripping column puts the


two streams in counter-current. The gas enters the column from the bottom
with a solute concentration of yn+1 and a flow rate of G. The liquid enters
at the top, flowing down the column by gravity, having an initial solute
concentration of x0 and flow rate of L.

This configuration has proved to most successful, since it doesnt require a


lot of energy for the flow transport. The two streams exchange matter along
the length of the column, before exiting on the opposite sides. In absorption, the solute moves from the gas to the liquid stream, while the opposite

55

happens in stripping.
The gas leaves the column at the top with a solute concentration of y1 and
the same flow rate G, while the liquid leaves at the bottom with a solute
concentration xn and the same flow rate L. If the two streams exchange
solute, for the case of absorption, it holds that y1 << yn+1 and x0 << xn .
The solute has moved from the gas to the liquid stream. Since the mass
transfer between the two phases does not occur instantaneously, but rather
bit by bit along the length of the column, the whole process can be viewed
as a cascade of separated steps, so-called stages, going from 1 to n. Mass
transfer takes place in every one of them.

The thermodynamic conditions (T and p) have an effect on the equilibrium


between the two phases and therefore also control the mass transfer rate.
When mass transfer takes place, a certain amount of energy is generally
involved. If this energy is significant, the temperature of the liquid or gas
stream changes. The process is not isothermal anymore, making the modelling of the operation much more complicated. Fortunately, in most of the
cases, we can neglect the temperature change effect, allowing us to design a
simpler process.

4.2
4.2.1

Gas plant
Solvent selection

A solvent is selected based on the following criteria:


a) Absorption capacity For a given degree of separation, the required
amount of solvent depends on the absorption capacity of the solvent.
High capacity means that the amount of required solvent is low and
vice versa. When the solvent reacts with the solute (chemical absorption), the capacity of absorption is high, but this reaction must be
56

reversible (under certain conditions such as at an elevated temperature). Otherwise the solvent cannot be regenerated.
b) Selectivity The selectivity is an important factor when the process
deals with the separation of one or more components from a mixture
of gases. Only when the solvent has more affinity for the components
to be separated, the process can be successful.
c) Solvent regeneration process Since the solvent is usually recycled,
regeneration should be taken into account: The relative ease of regeneration may determine the selection of the solvent.
4.2.2

Industrial schemes

If a gas stream has been purified by a suitable solvent in an absorption column, what happens to the polluted solvent? How expensive would it be to
use fresh solvent in every feed?
An appropriate approach to this would be regeneration of the liquid solvent
and recycle it to the absorption column. This is done by a stripping operation. Therefore, most of the time an absorption operation is coupled with a
stripping counterpart.
Different plant configurations are possible depending on the nature and difficulty of the solvent regeneration. Thermodynamic aspects (desorption temperature or decomposition temperature of the solvent), solvent volatility,
and chemical/physical aspects like corrosivity, viscosity, toxicity, as well as
costs, are taken into account when choosing the solvent for a specific absorption process keeping regeneration in mind. When solvent volatility is
very low, i.e. solvent is not present in the gas stream, a simple regeneration
process by heating is adequate. The system then looks very simple, and
following are three possible configurations:

57

In the first figure, the liquid absorbent to be recycled enters the stripping
column at the top and comes into contact with the stripping vapour. This
can be a steam or any inert gas with the right thermodynamic conditions
to strip the polluted solvent. The clean absorbent is then recycled to the
absorption column.

This second option has the to be recycled absorbent enter a re-boiled stripping column. The stripping vapour consists of the liquid solvent itself. Part
of the recovered solvent is continuously recycled to the absorber.

58

Lastly, a distillation column can also be used for regeneration. The polluted
absorbent is fed to the distillation column. In the bottom, the solvent is
collected and sent back to the absorber.

4.3
4.3.1

Absorption in a single equilibrium stage


Assumptions

In order to approach the study of the absorption/stripping operation, a


certain number of assumptions are needed in order to simplify the design
and easily understand the basics concepts:

In reality, most of the time, the above assumptions about gas and liquid
streams are true. However different options can be considered for the so59

lute, the equilibrium, and the temperature, i.e. multicomponent, nonlinear


equilibrium, nonisothermal operation.
4.3.2

The liquid-vapour equilibrium

Linear equilibrium
Typically in an absorption or stripping problem the solute, which has
to be removed, is present in the liquid or in gas phase at very low
concentration (<1% ). Henrys law is therefore used to represent the
equilibrium correlation for a solute A between the gas and the liquid
phase
yA = mxA
(115)
where
m=

H
Ptot

(116)

In the x vs. y diagram, the correlation can be represented as follows

Non-linear equilibrium
in the range of concentration when Henrys law can not be applied
anymore (xA >1% ), the relationship in between y and x is not linear
any more but generally curved, expressed as
yA = f (xA )
which can be visualised as:

60

(117)

Mole ratios vs. mole fractions in the equilibrium representation


Besides the mole fraction, there are others ways to express the gas and
the liquid composition. In some cases it can be more useful to express
it with the mole ratios, which refer to the moles of carrier instead of
the total number of moles as a reference:
Ni,liquid
Nn,liquid
Ni,gas
Yi =
Nn,gas

Xi =

moles of solute i in the liquid


moles pure carrier liquid (component n)
moles of solute i in the gas
=
moles pure carrier gas (component n)
=

(118)
(119)

Molar fractions and molar ratios are linked with the following relationships:
Xi
1 + Xi
xi
Xi =
1 xi

Yi
1 + Yi
yi
Yi =
1 yi

xi =

yi =

(120)
(121)
(122)

If the equilibrium is a straight line y = mx in the x vs. y diagram, it


will be a curve Y = f (X) in the X vs. Y counterpart:

61

However in most of the cases, as we have already pointed out, we are in


the condition of infinite dilution of the solute (very low concentration).
In this case the following simplification can be made:
xi Xi

yi Yi

(123)

which allows us to assume a linear equilibrium in the x-y and X-Y


plane.
As we can see, this is of a key importance to simplify the application of
the McCabe-Thiele graphical methods for the design of an absorption
or a stripping process.
4.3.3

Single equilibrium stage

During the absorption operation, the gas phase and the liquid phase must
be in contact. Before considering the different possible configurations, we
consider the thermodynamical aspects of this contact between phases.
The solute contained in G transfers to the liquid phase L. The concentration
in the gas decreases while the concentration in the liquid increases. The
pairs of points (concentration of solute in the gas and liquid phases) at each
moment constitute the operating line.

62

Mass transfer ends when equilibrium is reached. Graphically, it means that


the operating line crosses the equilibrium line. The driving force for absorption can be qualitatively seen as the distance between the equilibrium and
operating line. Therefore, when the two lines cross, the driving force is zero.
The gas at the inlet has a certain concentration of solute, represented by the
molar fraction, y0 . The solvent is put in contact with the gas phase. The
solvent can be pure or can have a certain concentration of solute, represented
by x0 .

The equilibrium between phases can be represented in an x vs. y diagram for


a given temperature and pressure. Sometimes the equilibrium is a straight
line, y = mx. The inlet compositions of the two phases can be represented
in the diagram.

63

After contact of the inlet gas stream with the solvent, the concentration of
the solute in the gas phase decreases: y < y0 . Likewise, the solvent is now
charged with solute: x > x0 . Because this is an ideal equilibrium stage,
equilibrium between the phases is reached. The mass balance equation is:

L x0 + G y0 = L x + G y

(124)

total mass of solute at inlet = total mass of solute at outlet


This can be solved with respect to y:


L
L
y = y0 + x0 x
G
G

(125)

L
The slope of the operating line is then G
:

Because this is an equilibrium stage, the equilibrium is reached and the


operating line touches the equilibrium line:

64

The equilibrium line represents a barrier and determines a region of values


L
(x, y) that cannot be reached for any ratio G
.
Using the equilibrium and operating line, a value for y corresponding to the
intersection can be found:


L
L
y = mx
(126)
y = y0 + x0 x
G
G



L
L y
y = y0 + x0
G
G m


L y
L m x0

y = y0 +
G
m
G m

(127)
(128)

Per definition, the ratio


A=

L
Gm

is called absorption factor A.




L y
y = y0 + A m x 0
G m

(129)

(130)

Using the equilibrium equation, the product m x0 can be replaced by y0 ,


i.e. the composition of a gas at equilibrium with a liquid of composition x0
(we do this, because we want to get all our xs out of the equation. The
usual equilibrium, y0 = mx0 doesnt work here, because the two variables
come from different streams).


L y
y = y0 + A y0
(131)
G m
65

This finally gives for y:

y0 + A y0
1+A
The fraction of absorption is defined as:
=

y=

(132)

1
y0 y
=1
y0 y0
1+A

(133)

Here, the numerator is the distance between y0 and y, while the denominator
is the distance between y0 and the composition of a gas at equilibrium
with x0 . This represents the maximum solute exchange. The fraction of
absorption can also be written in terms of the absorption factor A.
A0

y y0

y y0

y0 represents the minimum concentration of solute in the gas phase, so the


maximum absorption (A tends to infinity).
4.3.4

Co-current cascade configuration

The assumptions remain the same as for the last case. When the gas flow is
parallel to the solvent flow in a cascade of ideal stages, only the first stage
is effective. As we saw when we studied the single ideal stage, the solvent
flow is at equilibrium with the gas at the outlet. This means that the concentration of solute in the solvent is maximal and, at the same conditions
of temperature and pressure, the absorption operation cannot be more effective. There is no driving force at equilibrium.

66

A cascade of single stages, all of them at the same conditions, is then equivalent to one single stage.

First, we consider two ideal stages. The solvent is put in contact with the
gas phase in the first stage. The solvent can be pure or can have a certain
concentration of solute, represented by x0 .
The equilibrium between phases can be represented in an x-y diagram for
a given temperature and pressure. Again, lets consider it a straight line
y = mx.

The inlet compositions of the two phases can be represented in the diagram
(x0 , y0 ).
The content of solute in the gas after the first stage is now lower than at
the entrance: y1 < y0 , and the solvent is now charged with solute, x1 > x0 .
Because this is an ideal equilibrium stage, the equilibrium between the two
phases is reached.
Repeating what we saw before in the case of a single ideal stage, the change
in composition in the two phases can be represented in the following diagram:

67

L
The equation for y from before, with the operating line G
:


L
L
y = y0 + x0 x
G
G

(134)

For the single ideal stage, we derived the equation for y1


y=

y0 + Ay0
1+A

(135)

with A being the absorption factor. Now, we have a second ideal stage,
which can be handled analogously, following the same derivation as for the
single equilibrium stage:
y1 + Ay1
y2 =
(136)
1+A

68

or, for any stage:

yi1 + Ay0
(137)
1+A
The aim of a second stage would be to have y2 smaller than y1 , so that we
improve the operation by decreasing the concentration of pollutant in the
gas stream.
yi =

But since the equilibrium represents a barrier and determines a region of


L
values (x, y) that cannot be reached for any ratio G
and any number of
stages

we have already reached equilibrium and can move no further.


One can also arrive at this conclusion mathematically: Starting at
y2 =

y1 + Ay1
1+A

(138)

we reach equilibrium
y1 = y1

(139)

so we get
y2 =

y1 (1 + A)
(1 + A)

(140)

and finally
y2 = y1

(141)

The composition equation for the last stage mirrors the one for the first
yn =

y0 + Ay0
1+A
69

(142)

as does the fractional absorption:


=
4.3.5

yn+1 y1
1
=1

yn+1 y0
1+A

(143)

Cross-current cascade configuration

Again, the assumptions remain unchanged.


In this case the solvent stream is divided in n streams before entering the
n stages of the cascade. The solvent flows in cross-current configuration to
the gas flow.
At every stage the solvent that comes in is fresh. The solvent coming out is
at equilibrium with the gas stream and is not driven to another stage but
collected and mixed with the other solvent outlets.
That way, the performance of the operation is improved with respect to a
single stage or a co-current cascade.

Each step has a different operating line, since the conditions of the concentration in the gas change. The slope of the operating line is constant, since
we consider variations on the gas flow not being significant and the amount
of solvent is kept constant.
We consider three ideal stages.

70

The solvent stream is divided and each new stream is driven to a stage. The
solvent can be pure or can have a certain concentration of solute, represented
by x0 . Every stage is receiving solvent with the same composition x0 . Every
stage also receives the same amount of solvent, (L/3). The equilibrium
between phases can again be represented in an x-y diagram, for a given
temperature and pressure, as a straight line y = mx.

Again, the inlet compositions can be represented in the diagram; the content
of solute in gas is lower than at the entrance, while the concentration of
solute in the liquid is higher. The equilibrium between phases is also reached.
Compared to what we saw before, there is a subtle change in the expression
for y:


L/3
L/3
x0
x
(144)
y = y0 +
G
G
L
Therefore, the slope of our operating line is now ( 3G
), and, in terms of
absorption factor A:
y0 + (A/3)y0
y1 =
(145)
1 + (A/3)

71

Looking at the second ideal stage, the gas at the entrance has a concentration equal to y1 . Because the solvent has the same composition as before,
the new operating point must be located on the same vertical.
The outlet of the second stage is again at equilibrium, and because gas and
solvent flow are approximately constant, the new operating line has the same
L
slope: ( 3G
).

Since the equilibrium is reached, we can easily draw the new operating line.
For the concentration in terms of A, we get (material balance, divide by G,
y
L
replace (x2 = ym2 ), replace with A, replace (x0 = m0 ), replace Gm
with A,
solve for y2 ):
y1 + (A/3)y0
(146)
y2 =
1 + (A/3)
Other stages work analogously. We find y0 in the numerator for every stage,
since the fed solvent always has the same concentration x0 .

72

Sometimes, the solvent flows are chosen to be different at each stage. This
makes sense, since the concentration of solute in the gas decreases in each
stage and thus the driving force decreases as well as the solvent requirements.
Then, the slopes of the operating lines are different and must be calculated.

L1
G

L2
G

L2
G

(147)

In principle, the solvent flow rate will increase and thus the absolute value
of the slope will also increase.

Looking at the expressions for the final composition of the gas coming out
of the last stage (yn ), which is of primary interest, we see that we have to
solve the equation:
yi1 + (A/N )y0
yi =
(148)
1 + (A/N )
This can be rearranged to give
yi =

1
(A/N )
yi1 +
y
1 + (A/N )
1 + (A/N ) 0

(149)

which is a First Order Difference Equation. A First Order Difference equation of the form
yi = yi1 +
(150)
has the solution
yi = y0 i +

1 i
1

(151)

We see, that in our case the coefficients are


= y0

(A/N )
1 + (A/N )

73

1
1 + (A/N )

(152)

and the general solution therefore


yi = y0



1
1

+
y
1

0
(1 + A/N )i
(1 + A/N )i

(153)

So, for the last stage, where i = n:


yn = y0



1
1

+
y
1

0
(1 + A/N )n
(1 + A/N )n

(154)

and for the fractional absorption:


=
4.3.6

y0 yn
1
=1

y0 y0
(1 + A/N )n

(155)

Counter-current cascade configuration

Once again, the assumptions remain the same.


In the counter-current configuration, the gas flows in the opposite direction
to the solvent. The gas, rich in pollutant at the entrance, is put in contact
with the solvent already charged with pollutant (outlet). On the contrary,
on the other side, the gas now poor in pollutant is put in contact with the
clean solvent.
This is the characteristic of any counter-current process, and is convenient
because the driving force is nearly constant along the cascade (the operation). On the diagram the driving force is represented qualitatively by the
distance between the equilibrium and the operating lines.
Consider three ideal stages. The solvent stream enters the cascade by the
first stage. It can be pure or can have a certain concentration of solute,
represented by x0 . The gas enters the cascade by the last stage and flows in
counter-current to the solvent flow.

74

The operating line is defined by points of compositions of gas and liquid


phases at the same level of the process. Points on the operating line are:
(x0 , y1 )

(x1 , y2 )

(x2 , y3 )

(x3 , y4 )

Because the stages are ideal, the equilibrium is reached in each of them.
This mean that the gas and the liquid phase going out from a stage are at
equilibrium, and thus lie on the equilibrium line.
(x1 , y1 )

(x2 , y2 )

(x3 , y3 )

We now do a local (dotted area in the next figure) partial (only for the
solute) mass balance:

75

Gyj+1 + Lx0 = Gy1 + Lxj

(156)

Again, with the equilibrium equation y = mx and after dividing by G, we


get the operating line
yj+1 =
which gives us a positive slope

L
L
xj + ( x0 + y1 )
G
G
L
G

(157)

for our operating line.

Known data in this case are the initial gas composition, yn+1 , as well as the
initial solvent composition, x0 . When the final gas composition, y1 , is known,
because it is restricted to less than a certain minimum value (specification),
the first point of the operating line can be set.

76

L
For the specific slope ( G
), we can set the operating line through that point.

The graphical resolution consists of a step by step construction. We know


that x1 is at equilibrium with y1 . This lets us set x1 on the diagram. We
also know that x1 and y2 are on the same operating line, since they are at
the same levelin the cascade. And so on and so forth, until the specification
is reached.

77

A sequential resolution is also possible without the graphical representation,


alternating the use of the operating line equation and the equilibrium equation.
Introducing the absorption factor in the mass balance equation obtained
before gives us
yj+1 = Ayj + (y1 Ay0 )
(158)
Comparing this equation to the general Difference Partial equation, we get
= y1 Ay0

=A

Applying the general solution, we find




y1 Ay0
y1 Ay0

yj = y0
Aj +
1A
1A

A 6= 1

(159)

(160)

This can be transformed (j = n + 1) to


yn+1 = y0 An+1 + (y1 Ay0 )

1 An+1
1A

(161)

for the (n + 1)th stage.


The fractional absorption can now also be calculated by substituting the
value of yn+1 :
=

yn+1 y1
A1
= 1 n+1
yn+1 y0
A
1

A 6= 1

(162)

This is the Kremser equation. Solving the Kremser equation for n, the
number of stages to the specified purity can be calculated:


ln 1/A
1
n=
(163)
ln A
78

When A = 1, the Kremser equation cannot be used. The fraction of absorption is then calculated via
=

4.4
4.4.1

yn+1 y1
1
=1

yn+1 y0
1+A

A = 1

(164)

Absorption operations in tray columns


Mass balance

Lets consider an absorption case. In the most typical configuration the gas
and the liquid enter and leave the column at opposite ends (counter-current
configuration).

The gas enters from the bottom with a flow rate of G and a composition
of yn+1 , while the liquid enters from the top with a flow rate of L and a
composition of x0 .
The gas leaves the column at the top with the same flow rate of G and a
composition of y1 , while the liquid leaves the column at the bottom with
the same flow rate of L and a composition of xn .

79

This gives us the known overall mass balance:


Lx0 + Gyn+1 = Lxn + Gy1

(165)

For the mass balance for a section between stage 1 and the generic stage i,
we can write:
Lx0 + Gyi+1 = Lxi + Gy1
(166)

Solved for yi+1 , this equation gives:


yi+1 =
4.4.2

L
L
xi + (y1 x0 )
G
G

(167)

The operating line

The operating line is the points of compositions (xi , yi+1 ), where xi and
yi+1 are respectively the compositions of liquid and gas stream exchanging
matter in the generic stage i, as shown in the figure below:

(xi , yi+1 ) are correlated by the mass balance at the stage (i + 1):
yi+1 =

L
L
xi + (y1 x0 )
G
G
80

(168)

In the x-y diagram this correlation can be represented by a straight line with
L
slope G
and it is known as operating line.
L
In general the ratio G
is not constant since during the process the solute
moves from one phase to another, hence modifying the flow rates of both
the liquid and the gas phase.

However in the condition of infinite dilution of the solute (very low concentration), the amount of absorbed solute is so small that its passage from one
stream to the other doesnt significantly impact the flow rates of the two
involved streams.
Moreover the following assumptions have been made:
1. Solvent is not volatile
2. Carrier gas is not soluble
Therefore the liquid and gas flow rates can be considered constant and the
equation is represented in the x-y plan as a straight line:

If for any reason, the assumption of infinite dilution is not true, only the inert
part of the liquid and the gas phase can be considered and their compositions
would be expressed as mole ratios. In this case it is:
1. L0 = inert part of liquid (solvent without solute)
2. G0 = carrier gas (without solute)
3. Xi = liquid composition in mole ratio =
4. Yi = gas composition in mole ratio =

81

Gyi
G0

Lxi
L0

From this, we get mass balance


L0 X0 + G0 Yi+1 = L0 Xi + G0 Y1

(169)

from which we get the operating line


Yi+1 =

L0
L0
Xi + (Y1
X0 )
G0
G0

(170)

which is again a straight line in the X - Y plane.


More in general the mole ratio is not very often used because of the difficulty
of representing the equilibrium with a linear correlation in the (X-Y) plane.
Whenever it would be possible the assumption of very diluted solution will
be made, allowing for the use of the mole fractions to express the gas and
liquid compositions.
4.4.3

Problem description

A gas phase containing a pollutant with a concentration of yn+1 must be


purified. To this aim, a suitable liquid solvent is used. The initial liquid
solute concentration x0 is very low or zero.
The two streams exchange solute during the passage in the column.

The outlet gas is now cleaned and contains a pollutant concentration within
specification y1 .

82

The pollutant is removed with the outlet liquid which now contains a concentration of pollutant xn .
The column under examination is a rather standard tray column, where the
liquid and gas stream are brought into counter-current contact.
The operating conditions of the column, like temperature T and pressure P ,
are selected in order to favour the absorption process in terms of thermodynamic equilibrium, and to optimise the separation process.
Once again, the following assumptions are made:

The assumption, whether a linear or non-linear equilibrium is assumed, depends on the solute concentration in the liquid feed, xsolute . If it holds,
that xsolute < 1%, Henrys law can be applied and we can assume a linear
equilibrium:
H
y= x
(171)
P
For higher solute concentrations, the non-linear relationship y = f (x) between x and y has to be taken into account.
4.4.4

Process design and verification

Usually, when designing an absorption process, an engineer could face two


kinds of problems:
1. The column doesnt exist. The column must be sized in order to
perform the required operation.
2. The column already exists. It needs to be determined whether the
required purification can be performed with the existing column.

83

In the first case, we deal with a problem of column design, while in the
second case it is a column verification.
Either way, we know a couple of things before we start:
T ... Temperature
p ... Pressure
y = mx or y = f (x) ... Equilibrium data
G ... Gas flow rate
yn+1 ... Initial gas composition
x0 ... Initial Solvent composition
Then, depending on our task, the set up looks different:
1. Column design
In the case of column design, our specification is the final solute concentration in the gas stream, y1 .
No other data is needed.
Our remaining unknowns are:
(a) N ... Number of stages
(b) L ... Liquid flow rate
(c) xn ... Final concentration of the solute in the liquid outlet
2. Column verification
In the case of column verification, our specification is the number of
stages in the existing column, N .
Additionally, the liquid flow rate L is needed.
Our remaining unknowns are:
(a) y1 ... Final solute concentration in the outlet gas
(b) xn ... Final solute concentration in the outlet liquid

84

4.4.5

Process design - linear case

An absorption problem is usually presented as follows: There is a polluted


gas stream coming out of a process. The pollutant must be recovered in
order to clean the gas or because its a valuable compound. In general, the
conditions of the gas stream are known, the compositions can be measured,
and the gas flow rate can be set.

The temperature and the pressure at which the process takes place must be
chosen. The thermodynamics of the process at different conditions must be
studied in order to decide the most convenient operating conditions. The
solvent can be used pure or can come from a recycle, and thus contain some
pollutant. The initial composition of the solvent (x0 ) must be obtained by
analytical methods, or can be set by mixing pure and recycled solvent.
Once temperature, pressure and initial composition of the solvent are set we
can consider them as data. This is a list of known variables:

85

We have already seen in the counter-current cascade of ideal stages that a


graphical representation can help to visualize and solve the problem. Considering the initial compositions of the gas and solvent flows and the specification made for the gas outlet composition, we can draw in the diagram:

Now, there are infinite number of possible segments passing by the point
(x0 , y1 ) and ending in a point with y coordinate yn+1 . The operating line is
the equation of all the lines fulfilling:
yn+1 =

L
L
xn + ( x0 + y1 )
G
G

(172)

L
The slope of the operating line depends on the solvent flow rate ( G
). When
one of the lines is chosen as operating line, then the slope is set and thus,
the required solvent flow-rate can be calculated.

86

There is a special operating line among the infinite number we can choose.
This is the one that touches the equilibrium line at the point (yn+1 /m, yn+1 ).
The mass exchange is at its maximum because the equilibrium is reached,
but the required number of stages is infinite.
 
L
1
(173)
Amin =
G min m
The slope corresponding to this line is the smallest possible slope, because
the process cannot go beyond he equilibrium line. The minimum slope can
also be written in terms of the fractional absorption for a linear equilibrium:
 
L
yn+1 y1
yn+1 y1
m(yn+1 y1 )
=
=
=
= m
(174)
G min
xn x0
yn+1 /m x0
yn+1 y0
In practice, the slope for the operating line is taken as 1 to 2 times the
minimum slope:
 
L
L
= f actor
f actor (1, 2)
(175)
G
G min
Once the operating line is set, we can proceed with the construction seen in
the counter-current stage configuration. This gives us the number of stages,
n.

We already know all the initial unknowns: L, n and xn .

87

The next two concepts are very important. They can be reviewed in the
ideal equilibrium stage chapter.
1. Absorption factor
L
Gm

(176)

y0 y1
y0 y0

(177)

A=
2. Fraction of absorption
=

As we saw before, in the case of a counter-current configuration, the fraction


of absorption can also be written in terms of the absorption factor, A. This
gives us the Kremser equation for the case in which A is not equal to 1:
A1
An+1 1
1
=1
1+A
=1

A 6= 1

(178)

A = 1

(179)

Therefore, the number of stages required, n, can be calculated using the


two forms: the fraction of absorption definition and the Kremser equation.
The number of stages obtained by this method should be equal to the one
obtained by the graphical construction.
4.4.6

Process design - nonlinear case

Up to now, we always treated the case of linear equilibrium. In the linear


case, the equilibrium can be described with the simple equation y = mx.
However, we dont always have such an easy relation between the gas and the
liquid phase. The equilibrium relation can only be described by the general
formula:
y = f (x)
(180)

88

When y = mx does not apply, Kremsers equation is useless. However, the


graphical stage-by-stage construction (and the corresponding calculation)
can still be applied.
Considering the initial compositions of the gas yn+1 and of the solvent x0
and the specification made for the gas outlet composition y1 , we can draw
in the diagram:

The operating line is not influenced by a nonlinear equilibrium. Therefore,


the same operating line equation is still valid.
L
L
xn + ( x0 + y1 )
(181)
G
G
The slope of the operating line depends on the solvent flow-rate L. When
one of the lines is chosen as operating line, then the slope is set and thus,
the required solvent flow-rate can be calculated.
yn+1 =

There is a special operating line among the infinite number we can choose.
This is the one that touches the equilibrium line at the point (xn , yn+1 ).

89

Therefore, the following equation is valid at that point:


yn+1 = f (xn )

(182)

The slope corresponding to this line is the smallest possible slope, because
the process cannot go beyond the equilibrium line.

The slope of this operating line can be calculated:


 
yn+1 y1
L
=
G min
xn x0

(183)

In practice, the slope for the operating line is taken as 1 to 2 times the
minimum slope:
 
L
L
f actor (1, 2)
(184)
= f actor
G
G min
Once the operating line is set, we can proceed with the construction seen in
the linear case. The usual stage-by-stage construction is adopted, switching
between equilibrium line and operating line. This gives us the number of
stages, n.

90

In the beginning, we had the following unknowns: L, n and xn . The liquid


L
flowrate L can be calculated with the slope of the operating line G
, knowing
the gas flow rate G. The number of stages n and the mole fraction xn can
be read out of the graph. In our example the stage number is at least n = 3.
4.4.7

Process verification - linear case

In this case we have a polluted gas current that we would like to clean.
On the other hand we have an absorption equipment, and we would like to
evaluate whether the operation can be performed with this equipment and
under which conditions.
The temperature and the pressure at which the process takes place must be
chosen. The thermodynamics of the process at different conditions must be
studied in order to decide the most convenient operating conditions. The
solvent can be used pure or can come from a recycle, and thus contain some
pollutant.
The initial composition of the solvent (x0 ) must be obtained by analytical
methods, or can be set by mixing pure and recycled solvent.
The geometry of the equipment is well known.

91

As we have seen before, the problem can be represented in the x-y plane.
The equilibrium line can be drawn to start.
Considering the initial compositions of the gas and solvent flows and the
specification made for the gas outlet composition, we can draw in the diagram:

92

Because we know both of the molar flow rates, the absorption factor is fixed,
and so the slope of the operating line is known:

A=

L
Gm

slope =

L
G

(185)

L
There are an infinite number of lines with the slope G
that pass through a
point of y-coordinate yn+1 and a point along the x-coordinate x0 .

Given the fact, that we have a set number of stages (of our already existing
equipment), there is now only one operating line that fulfils the condition
yn+1 =

L
L
xn + ( x0 + y1 )
G
G

(186)

The procedure is the following: the operating line is moved up and down,
until the number of stages obtained coincides with the number of stages that
the equipment provides. On the diagram, we see how by moving the operating line, we go from one to six stages. If the available equipment provides
six stages, we have found the right operating line in the second case.

93

When the line is fixed, the final concentration of the gas is obtained. This
concentration can be then evaluated in order to clarify whether the process
has for instance, the desired performance. The outlet composition of the
solvent is also obtained.
4.4.8

Process verification - nonlinear case

Again, we have a polluted gas current that we would like to clean as well as
preexisting absorption equipment, and would like to evaluate whether the
operation can be performed with this equipment and under which conditions.
The temperature and the pressure at which the process takes place must be
chosen. The thermodynamics of the process at different conditions must be
studied in order to decide the most convenient. The solvent can be used
pure or can come from a recycle, and thus contain some pollutant.
The initial composition of the solvent (x0 ) must be obtained by analytical
methods, or can be set by mixing pure and recycled solvent.
The geometry of the equipment is well known. Therefore the number of
stages n is set and cannot be changed.

94

Once temperature, pressure and initial composition of the solvent are set we
can consider them as data.

As we have seen before, the problem can be represented in the x-y plane.
The equilibrium represented by the function y = f (x) can be drawn to start.
Considering the initial composition of the gas yn+1 and the initial composition of the solvent x0 , we can draw in the diagram:

95

Because we know the value of the solvent flow rate L and the gas flow rate
L
G, the slope of the operating line G
is known.
L
pass by a point
There are infinite number of lines that having the slope G
of y-coordinate yn+1 and by some point with x-coordinate x0 .

Again, given the fact, that we have a set number of stages (of our already
existing equipment), there is now only one operating line that fulfils the
condition
L
L
yn+1 = xn + ( x0 + y1 )
(187)
G
G
The procedure is the following: the operating line is moved up and down,
until the number of stages obtained coincides with the number of stages
that the equipment provides. On the diagram, we see how by moving the
operating line, we go from one to four stages. If the available equipment
provides three stages, we have found the right operating line.

96

When the line is fixed, the final concentration of the gas y1 is obtained. This
concentration can be then evaluated in order to clarify whether the process
has the desired performance, for instance. The outlet composition of the
solvent xn is also fixed.

4.5
4.5.1

Stripping operations in tray columns


Comparison: absorption vs. stripping

Absorption and stripping are two very similar processes. As you can see
from the table below, the principles are the same but the direction of the
mass transfer is the one the opposite of the other.

97

Absorption

Stripping

A gas stream is purified using a


liquid solvent.

A liquid stream is purified using a


gas stream.

Streams at stage i:

Streams at stage i:

x-y diagram (mole fraction):

x-y diagram (mole fraction):

4.5.2

Process design

Stripping processes can be an alternative to distillation if the liquid stream


is not stable at the required distillation temperature, or the solute has a
boiling point similar to the one of the other components present in the liquid stream or finally when the solute is present at infinite dilution condition
98

in the liquid stream.

The process of finding the unknown data is very similar to the absorption
case. First, we can draw the linear equilibrium line y = mx. Apart from
that, we have xn , our target pollutant concentration specification in the liquid stream, have x0 , our initial pollutant concentration in the liquid stream,
and know yn+1 , our initial gas stream, which can be pure or already contain
some pollutant from recycling.
Through our known starting point (xn , yn+1 ), an infinite number of operating lines can be drawn. From the material balance, we get the usual
yn+1 =

L
L
xn + ( x0 + y1 )
G
G

(188)


L
There is one operating line with a maximum slope G
, though, which
max
goes through the point in which the equilibrium line has the x-coordinate
of x0 . This maximum operating line unfortunately also requires an infinite
number of stages.

99

In practice, the the maximum slope of the operating line fulfils the following
condition
L
1

=
(189)
L
G
a G max
or:
G
=a
L

L
G


with a (1.2; 2)

(190)

max

Now that we have our operating line, G and n can be constructed as usual.

For the the Kremser equation for stripping instead of absorption we intro-

100

duce equivalents to and A: The stripping fraction,


x0 xn
=
x0 xn+1
with xn+1 =

yn+1
m ,

(191)

and the Stripping factor S


mG
L

(192)

S n+1 S
S n+1 1

(193)

S=
The Kremser equation is now:
=

Solving for n gives the number of stages (which, has to be rounded to the
next integer). Using this new integer n, we can go backwards and find the
exact stripping fraction of our system as well as the exact composition of
our liquid outlet stream, x0 ,
=

S n+1 S
x0 xn
==
n+1
S
1
x0 xn+1

(194)

The initial gas concentration y1 can be calculated via the mass balance from
earlier:
L
L
(195)
yn+1 = xn + ( x0 + y1 )
G
G
4.5.3

Process verification

As for the absorption case, in the stripping case under study here, the question is whether a certain operation can be performed with the given equipment or not. The problem is now to verify if the equipment is suitable for the
required stripping operation. This problem is handled like with the already
treated absorption case, with modified data, specifications and unknowns:

101

Concerning the nonideal case of a stripping problem, please refer also to the
described cases for absorption.

4.6

Nonideal operations

We consider the following assumptions:

In the reality, a tray does not work like an ideal equilibrium stage and the
ideal equilibrium compositions are very rarely reached in the real working
conditions.
For this reason the introduction of the efficiency concept is needed to take
into account the nonideal behavior of a tray.

The Murphree efficiencies are defined for both gas (E M G ) and liquid (E M L )
phase and for every tray of the column with the index i, as follows:

102

yi+1 yi
yi+1 yi
xi xi+1
=
xi xi+1

EM G =

(196)

EM L

(197)

However to simplify the column design or verification, we will always assume


that all the trays have the same value of Murphree efficiency.
The Murphree efficiency is typically a given data of the problem. In this way
the two expressions here above are two equations which constitue a kind of
pseudo-equilibrium line in the x-y plane.
The E M G equation is used to design absorption process while the E M L
equation is used for stripping problems.
To calculate the number of stages in a nonideal case, the McCabe-Thiele
method can be still applied but the Murphree efficiency line is used instead
of the ideal equilibrium line.
Please check the absorption case with efficiency (next section) how to apply
the McCabe-Thiele method using the Murphree efficiency line.
The overall efficiency E0 is defined as the ratio between the theoretical
number of equilibrium stages to the actual number of stages required:
Nequil.
(198)
Nactual
Even if the overall efficiency E0 is conceptually different from the Murphree
efficiency, most often it is assumed that both have the same value.
E0 =

4.6.1

Absorption with efficiency stage

So far, we always treated a case where the equilibrium is reached in every


stage. However, this assumption is often not correct. Therefore we need to
use an efficiency to describe the non-ideal behaviour.
The Murphree efficiency compares the actually absorbed amount in stage
i (yi+1 yi ) to the maximum possible amount absorbed in stage i if the
equilibrium was reached (yi+1 yi ).
We assume, that the Efficiency is constant during the whole column and
does not depend on the stage i. Therefore we get:
yi+1 yi
E = const. =
(199)
yi+1 yi
103

For the operating line we need the mass balance for the column from stage
1 to stage i:
L
L
xi + ( x0 + y1 )
(200)
G
G
The equation for the equilibrium is our third equation to solve the problem.
yi+1 =

yi = mxi

(201)

First we draw the equilibrium line and the operating line into the diagram

Now we need to draw a third line using the Murphree efficiency. This line
is drawn between equilibrium and operating line at a fractional vertical distance from the equilibrium line equal to the Murphree gas efficiency.
We get the line by combining the last three equations to:




L
L
yi = m + (1 E) xi + yi x0 (1 E)
G
G

(202)

This line replaces the equilibrium line in the graphical construction to determine the number of non-equilibrium stages.

104

Instead of the graphical solution, it is again possible to find the number of


stages mathematically.
L
Multiplying the mass balance with E
A and keeping in mind, that A = mG ,
we get
E
E
yi+1 = Emxi + y1 Emx0
(203)
A
A
We can combine the equilibrium condition yi = mxi and efficiency definition
to get
Emxi = yi yi+1 (1 E)
(204)

Combining the last two equations over Emxi we get:


E
E
yi+1 = yi (1 E)yi+1 + y1 Ey0
A
A
Solving for yi+1 :





A
E
yi+1 =
yi +
y1 Ay0
E + A EA
E + A EA
We can introduce the two new variables A0 and B 0




A
E
0
0
A =
B =
E + A EA
E + A EA

(205)

(206)

(207)

so we get the following equation:


yi+1 = A0 yi + B 0 (y1 Ay0 )
If we set the efficiency E = 1, we get
A0 = A

B0 = 1
105

(208)

As a next step, we would like to calculate the number of real stages. This
calculation needs some mathematical effort. Instead of our last result,
yi+1 = A0 yi + B 0 (y1 Ay0 )

(209)

we want an equation for i instead of i + 1.


We can see that the last equation is once again a First Order Difference
equation of the form
yi = yi1 +
(210)
with the general solution
yi = y0 i +

1 i
1

(211)

= B 0 (y1 Ay0 )

(212)

In our case, the coefficients are


= A0

Using the solution of the our Difference equation and the definition of the
efficiency E to get rid of the (i + 1)s, we get:


B0
B0

0 i
yi = y1 (1 E) + Ey0
(y

Ay
)
(y1 Ay0 ) (213)
1
0 (A ) +
0
1A
1 A0
The fractional absorption can now be calculated. After some mathematical
transformations, we finally get:
0

yn+1 y1
(A0 )n 1
=
yn+1 y0
(A0 )n0 A1

(214)

The number of real stages n is then:


n0 =

1
1 /A
log
log A0
1

(215)

The Kremser equation for the ideal case where equilibrium is reached on
every stage:
1 /A
1
log
(216)
n=
log A
1
Dividing the second equation by the first one, we get the overall efficiency
E0 :
n
log A0
E0 = 0 =
(217)
n
log A
It is important to note that the overall efficiency E0 and the Murphree
Efficiency E are not the same.
106

4.7
4.7.1

Multicomponent operations
Design of a multisolute absorption

In a real situation most likely the absorption or the stripping operation must
be designed to remove more that one solute, as so far considered in the simplified case.
In a multisolute absorption case, more than one compounds of the carrier
gas must be absorbed in the liquid solvent, while on the other way around,
in a multisolute stripping operation more than one liquid solutes move into
the gas phase.
The McCabe-Thiele and Kremser equation methods, so far learnt for one
simple solute, can still be applied to the multisolute case.
Among all the solutes, typically only one solute plays the role of key component. This is chosen in such a way that being the most difficult to remove,
the column size will end in the higest number of stages.
In this first phase of the multisolute column design, the task is to accomplish
the specification of the key component. The problem is then the design of
a column as it would be done for one single key solute.
An iterative procedure is implemented at the end of the design, to verify
whether the compositions of the other solutes are also below the given specification. If not, the choice of the key component was not the proper one
and another key component must be chosen in order to re-design the column.

107

It works as follows: Once temperature, pressure and initial composition of


the solvent are set we can consider them as data.

Up to this point we have been restricted to cases where there is a single solute
to recover. Both the stage-by-stage McCabe-Thiele procedure (for linear and
non-linear equilibrium) and the Kremser equation (for linear equilibrium)
can be used for multisolute absorption if certain assumptions are valid.
The single solute analysis by the Kremser equation requires
1. Systems that have a linear equilibrium
2. Systems that are isothermal
3. Systems that are isobaric
4. Systems that have negligible heat of absorption
108

5. Systems that have constant flow rates


These assumptions are still required. A new assumption must be added for
the multisolute case:
6. Solutes are independent of each other.
So the equilibrium for any solute does not depend on the amounts of other
solutes present. This assumption requires dilute solutions.
The consequence of assumption 6 is that we can solve the multisolute problem once for each solute, treating each problem as a single-solute problem.
This is true for the Kremser equation (assuming linear equilibrium) and for
the stage-by-stage solution method (any equilibrium relation).
Usually one outlet composition for one of the components is fixed. This
component is called key component and is the specification of the problem.
It is designed by the letter k.
The equilibrium information for each component is required, in order to
solve the problem. Lets consider the absorption problem of c components,
all of them systems of linear equilibrium:

Now, for the key component we can write the operating line as usual (note
that the slope of the operating line does not depend on the specific solute:

L
L
xk,n + xk,0 + yk,1
(218)
G
G
The problem can be solved for component k as usual, either using Kremser
or the graphical construction:
yk,n+1 =

109

k =

yk,n+1 yk,1
= Ak,min
yk,n+1 yk,0

(219)

Choosing A
Ak = factor Ak,min

(220)

Calculating the slope of the operating line:


L
= Ak m k
G

(221)

Using Kremser equation:


n=

1
log
log Ak

1 k /Ak
1 k


(222)

Once the number of stages has been found from the key solute, k, the concentrations of the other solutes can be determined by solving (c 1) fully
specified simulation problems.
This means that the number of stages is known and the outlet compositions
have to be calculated. This problem is similar to the one we have already
discussed and called simulation problem. The operating lines for every soL
lute have the same slope ( G
).
If we consider the next example, the number of stages found for the key component must be drawn for the other components. Graphically, this means

110

While analytically, the following steps are followed for all the components,
j, except for k

Aj =
j =

L
Gmj
An+1
Aj
j
An+1
1
j

(223)
(224)

So the outlet composition in the gas can be calculated:

yj,1 = yj,n+1 j (yj,n+1 yj,0


)

4.8
4.8.1

Non-isothermal operations
Design of a non-isothermal operation

In this case, we are operating under the following assumptions:

111

(225)

Sources of heat effects:


Heat of solution
Heat of vaporisation and condensation
Heat exchange between liquid and vapour phase
Heat exchange with the outside (open system)
In most absorption cases heat effects are only very small and can therefore
be neglected. But there are some cases, where heat is produced during
absorption. Some important examples where heat effects are important are
as follows:
Absorbing ammonia with water
Absorbing acids with water
Absorbing water with acids
Absorbing Acetone with water
As the solubility of gases in liquids decreases with temperature, the absorption factor gets smaller and absorption becomes less efficient. The easiest
way to solve such a problem is adding surface to the column to improve heat
transfer or adding a cooling circuit. If the heat transfer to the outside is
large enough, heat effects at the inside of the column can be neglected again.
The problem can also be solved mathematically. Following approaches ranging from very empirical to exact modelling are usually used:
Use of the isothermal approach and adding a design safety factor of
1.5 to 2 (especially for packed columns)
Adiabatic approach
Semi-theoretical short-cut methods
Simulation
This simple diagram shows the notation that will be used

112

The liquid enters the column at its top flowing by gravity trough the column. The gas enters at the bottom. With this configuration, work is not
required for the flow transport.
The heat of absorption of solute (gas) increases the temperature of the solution. When the gas (G) enters the column at about 10 to 20 C under
the liquid temperature going out (L), and the liquid is volatile, evaporation
will cool down the liquid at the bottom of the column. This means that the
temperature profile in the column will present a maximum near its bottom.
The following diagrams represent the temperature of the liquid across the
column.
Case 1: The heat of absorption is not significant and the effect of liquid
evaporation can be neglected.

113

When the heat of absorption is not significant and the effect of liquid
evaporation can be neglected, the adiabatic absorption column has a
constant temperature along the height.
Case 2: The heat of absorption cannot be neglected.

114

Along the column, the absorption of gas increases the temperature.


Gas and liquid temperatures are similar at the bottom, and the gas
entering the column is already saturated. The changes in temperature
are only due to the heat of absorption of solute gas. This is why
temperature only increases along the column.
Case 2: The heat of absorption cannot be neglected, the effect of liquid
evaporation cannot be neglected.

115

Along the column, the absorption of gas increases the temperature.


The gas enters the column at a lower temperature than the liquid and
thus, at the bottom the evaporation of liquid is important. This is
why the temperature decreases at the bottom.
We will now set up the balances for the adiabatic approach. In the adiabatic
approach energy balances are calculated under following assumptions:
There is no heat exchange through the walls of the column
The gas and liquid stream leaving the stage are at the same temperature
Therefore two new concepts become important:
1. Energy balances
2. Changes of L and G within the column
Calculating changes of L and G within the column are important when
dealing with large solute fractions. This is often the case when calculating
non-isothermal columns.

116

Once, pressure and initial composition and temperature of the solvent are
set we can consider them as data. This is a list of known variables:

The material balance over the column gives us


L0 + Gn+1 = Ln + G1

(226)

L0 x0 + Gn+1 yn+1 = Ln xn + G1 y1

(227)

y = f (x, T )

(228)

The energy balance over the column gives us


L0 HL0 + Gn+1 HGn+1 = Ln HLn + G1 HG1
HL =
HG =

HL0 + cp,L (T T 0 ) +
0
HG
+ cp,G (T T 0 )
117

(229)
HS

(230)
(231)

using the following symbols:

HS represents the molar enthalpy of mixing or the integral heat of solution, at the prevailing concentration and at the base temperature T 0 .
The amount of liquid increases from the bottom to the top of the column.
It is calculated by adding up the amount of solvent and the amount of absorbed solute. The amount of gas is calculated the same way. It decreases
from the top to the bottom of the column as the solute absorbs to the liquid.
To solve the problem, the gas outlet temperature and the amount of liquid solvent have to be estimated or calculated by another method, e.g. the
isothermal approach.
Now all necessary values of the bottom of the column can be calculated by
the energy and mass balances.
From the bottom of the column a new set of mass and material balances
can be made including the first stage:

From these balances all unknowns of the first stage can be calculated. This is
repeated including one more stage every time until the required specification
118

is reached.

4.9
4.9.1

Absorption operations in packed columns


Problem description

A packed column is a cylindrical vessel under pressure, filled with a suitable


packing material between the internal section 1 and 2, as shown in the
following figure

The liquid flows down by gravity through the surface of the packing elements
forming a thin film layer around it. The gas rises in counter-current mode
with the liquid stream.
The mass transfer takes place through the interface film present between
the gas and the liquid.
The column is now a continuous differential contacting device. There is no
distinguishable stage (like the plates in a tray column) present any more.
The column is needed for the same reason as previously examined absorption
units: a polluted gas stream, coming out from a process, must be cleaned
from the polluting compound (= solute) thanks to the use of a liquid solvent.
The objective is to design the suitable column to perform the required mass
transfer operation. In the case of packed column, instead of calculating the
number of stages, the design consists of determining the amount of packing
material needed in terms of meters of column height, H.

119

An increase of the height of the column, in terms of either number of plates


or meters of packing material, always has the effect of improving the column
mass-transfer performance since the contact time between the two exchanging phases is longer. Of course, other aspects, i.e. economics and pressure
drops, need to be accounted for. The right column height is the one that
compromises all the factors and fulfils the requirements at the lowest costs.
4.9.2

Design based on the HTU/NTU concept

Packed columns are continuous contacting devices that do not have the physically distinguishable stages found in trayed columns.
In practice, packed columns are often analysed on the basis of equivalent equilibrium stages using a Height Equivalent to a Theoretical Plate
(HET P ):
HET P =

packed height
number of equivalent equilibrium stages

(232)

Knowing the value of the HET P and the theoretical number of stages n of
a trayed column, we can easily calculate the height H of the column:
H = n HET P

(233)

The HET P concept, unfortunately, has no theoretical basis. HET P values can only be calculated using experimental data from laboratory or
commercial-size columns.
For packed columns, it is preferable to determine packed height from a more
theoretically based method using mass transfer coefficients.
The absorption problem is usually presented as follows. There is a polluted
gas stream coming out from a process. The pollutant must be recovered in
order to clean the gas.

120

At the bottom and the top of the column, the compositions of the entering
and leaving streams are:
(x1 , y1 )

(x2 , y2 )

(234)

Furthermore, we introduce the coordinate z, which describes the height of


the column. The green, upper envelope is needed for the operating line of
the absorption column.
First, we need a material balance around the green, upper envelope of the
column. It is the operating line, going through the point (x2 , y2 ):
Lx + Gy2 = Lx2 + Gy
L
y = (x x2 ) + y2
G

(235)
(236)

Additionally, we have the equilibrium condition:


y = mx

(237)

We can now draw the equilibrium and operating lines into the diagram.

121

L
From the operating line with the smallest slope ( G
)min , we can get
the known formula:
 
L
L
=f
with f (1, 2)
G
G min

L
G

with

(238)

As a third equation, we need a mass transfer rate equation. We take a small


slice of the column. The material balance over the gas side of this slice gives:

INgas = OU Tgas + OU Tmass transf er

(239)


mol s1

SGy(z) = SGy(z + z) + N aSz

(240)

S is the cross-sectional area of the tower. Please note that N , G and L are
defined as fluxes and not as molar flow rates [mol s1 ]:


molar flow rate
mol
G=
[G] =
(241)
column section S
cm2 s
mass transfer surface
a=
volume of the column

cm2
[a] =
cm3



(242)

Determination of the packed height of a column most commonly involves


the overall gas-phase coefficient Ky because the liquid usually has a strong
affinity for the solute. Its driving force is the mole fraction difference (yy ):


mol
N = Ky (y y )
[N ] =
(243)
m2 s
Dividing the mass transfer rate equation by S and z, we get:
Na = G

y(z + z) y(z)
z
122

(244)

Because we want a differential height of the slice, we let z 0:


Na = G

dy
dz

(245)

Introducing the definition of N :


G

dy
= Ky a(y y )
dz

(246)

Separating variables and integration gives:


Z

y2

dz =

H=
0

y1

G
dy
Ky a (y y )

(247)

Taking constant terms out of the integral and changing the integration limits:
Z H
Z y1
G
dy
H=
dz =
(248)
K
a
(y

y)
y
0
y2
The right-hand side can be written as the product of the two terms HOG
and NOG :
Z y1
G
dy
HOG =
NOG =

Ky a
y2 (y y )
H = HOG NOG

(249)

The term HOG is called the overall Height of a Transfer Unit (HTU) based
on the gas phase. Experimental data show that the HTU varies less with G
than with Ky a. The smaller the HTU, the more efficient the contacting.
The term NOG is called the overall Number of Transfer Units (NTU) based
on the gas phase. It represents the overall change in solute mole fraction
divided by the average mole fraction driving force. The larger the NTU, the
greater is the extent of contacting required.
Now we would like to solve the integral of NOG . We replace y with mx
using the equilibrium condition:
Z y1
dy
NOG =
(250)
y2 (y mx)
Solving the material balance for x, knowing that A =
in the previous equation:
Z y1
A dy
NOG =

y2 (A 1)y + y2 Ay2
123

L
Gm

and replacing x

(251)

After Integration and some transformation, we get:




A
1
(A 1) y1 y2
NOG =
ln
+
A1
A
A y2 y2

(252)

We already know the fraction of absorption :


=

absorbed amount
y1 y2
=
max. absorbed amount
y1 y2

(253)

Introducing and doing some transformations, we finally get for NOG :




A
1 /A
NOG =
(254)
ln
A1
1
The height of the column can be calculated in two ways:
H = HOG NOG = nHET P

(255)

The NTU and the HTU should not be confused with the HETP and the
number of theoretical equilibrium stages n, which can be calculated with
the Kremser Equation


1
1 /A
n=
ln
(256)
ln A
1
When the operating and equilibrium lines are not only straight but also
parallel, NTU= n and HTU= HET P . Otherwise, the NTU is greater than
or less than n.

When the operating and equilibrium lines are straight but not parallel
(NTU6= n), we need a formula to transform them. We can write:
HET P = HOG

NOG
n

(257)

Replacing NOG and n by the formulas found earlier, we get for HET P :
HET P = HOG

124

A ln A
A1

(258)

Doing the same calculation for NOG , we find:


NOG = n

A ln A
A1

(259)

Finally we want to calculate the volumetric overall mass transfer coefficient


Ky a. We know that
G
H
=
HOG =
(260)
NOG
Ky a
Solving for Ky a, we find:
GNOG
(261)
H
Now we want to focus on a stripping problem, which is usually presented as
follows: There is a polluted liquid stream coming out from a process. The
pollutant must be recovered in order to clean the liquid.
Ky a =

First, we need a material balance around the green envelope of the column.
It is the operating line, going through the point (x1 , y1 )(and also, again, the
equilibrium condition):
Gy1 + Lx = Gy + Lx1
L
y = (x x1 ) + y1
G
y = mx

(262)
(263)
(264)

We can now draw the equilibrium and operating line into the diagram. From
L
L
the operating line with the largest slope ( G
)max , we can get G
with the
known formula:
 
L
1 L
=
f (1.2; 2)
(265)
G
f G max

125

As a third equation, we need a mass transfer rate equation. We take a small


slice of the column. The material balance over the liquid side of this slice
gives:
INliq = OU Tliq + OU Tmass transf er
SLx(z + z) = SLx(z) + N aSz

(266)


mol s1

(267)

The flux N involves the overall liquid-phase coefficient Kx and the driving
force (x x ):
N = Kx (x x )
(268)
Dividing the mass transfer rate equation by S and z, we get:
L

x(z + z) x(z)
= Na
z

(269)

We let z 0 and introduce the definition of N :


L

dx
= Kx a(x x )
dz

Separating variables and integration gives:


Z H
Z x2
L
dx
H=
dz =
Kx a x1 x x
0
126

(270)

(271)

The term HOL is called the overall Height of a Transfer Unit (HTU) based
on the liquid phase. The term NOL is called the overall Number of Transfer
Units (NTU) based on the liquid phase.

HOL

L
=
Kx a

x2

NOL =
x1

dx
(x x )

H = HOL NOL

(272)

We already know the fraction of stripping :


=

amount stripped
x2 x1
=
max. amount strippable
x2 x1

(273)

Furthermore, we know the stripping factor S:


S=

mG
L

(274)

The solution of the integral of NOL can be found if one proceeds exactly as
in the case of absorption:


S
1
S 1 x2 x1
NOL =
ln
+
(275)
S1
S
S x1 x1
Finally, after some transformations, we find:


S
1 /S
NOL =
ln
S1
1

4.10
4.10.1

(276)

Chemical absorption
Principles

In chemical absorption, the gas to be removed dissolves and reacts with the
solvent and remains in solution. The reaction can be either reversible or
irreversible.
Example of the first case is the chemical absorption of gaseous CO2 from
air with aqueous amine solution:
CO2 + Amine = Product

(277)

Examples of the irreversible chemical absorptions are the reactions of acid


gases with base aqueous solutions:
HClair = NaOHaq. NaClaq.

127

(278)

HClair reacts with NaOHaq. , giving a salt, which precipitates, thus consuming solvent, which cannot be regenerated anymore.
For this reason, absorption with irreversible chemical reaction is most often used only in small facilities, where the absorber is usually small and no
regeneration facilities are required. In most of the cases, absorption with
reversible chemical reaction is preferred.
The following table presents the most common applications in industries for
chemical absorption.

In the following section, a case study is presented.


4.10.2

Case study: H2 S absorption with Diethanolamine

A natural gas stream containing methane (CH4 ) is discharged, contaminated


with Hydrogen Sulfide (H2 S). The H2 S can be removed with an absorption
operation.
Two factors have an important effect on the absorption task: the mass transfer velocity and the equilibrium between the two phases.
However, in the following example, our attention will mainly focus on the
equilibrium parameter, thus on the vapour pressure of the H2 S in the gas
128

phase as a determining factor for the absorption efficiency.


The higher the vapour pressure is, the smaller the amount of H2 S which
can be absorbed per m3 of solvent, resulting in a higher amount of solvent
requirement.
When pure water is used as solvent, every molecule of absorbed H2 S contributes to increase the vapour pressure in the gas phase. In the case of
Diethanolamin (DEA) is added in the water, part of the H2 S will dissociate
in HS , which will remain in the water and not determine any vapour pressure increase of the gas phase. Consequently, more H2 S can be absorbed in
the solvent phase.
In the DEA solvent the main part of the H2 S is dissociated, while the pH
value of the DEA is below 10. The following three reactions take place in
the solvent (capital letters denote concentrations):

*
H2 S + H 2 O )

HS + H3 O

W +S
KS =
= 0.683 107
S

(279)

*
(HOCH2 CH2 )2 NH + H2 O )

(HOCH2 CH2 )2 NH2 + OH

(280)



E+W
5
KS =
= 2.509 10
E

*
2 H2 O )

H3 O + OH

(281)



+

14
KW = W W = 0.741 10
Now we can also write the following:
W + + E = W + S

(282)

M = E + E = total concentration of DEA


+

B = S + S = total concentration of absorbed H2 S


129

(283)
(284)

B is also dependent on the DEA concentration with a factor,


B = M

(285)

Now, assuming that


W << S

W + << E +

S << S

(286)

the concentration of S is negligible and we can write:


S = B

E+ = B

(287)

from the previous equations, substituting and arranging:


W = KE

M B
B

W+ =

B
KW
KE M B

S=

B2
KW
KE KS M B

(288)

the molar fraction of H2 S can then be calculated:


S
ctot
B2
KW
=
KE KS ctot M B
KW M 2
=
KE KS ctot 1

(289)

ctot xH2 S
B
=
M
M

(290)

xH2 S =

But
=
so that

xH2 S

xH S
KW
2
=
KE KS xM xH2 S

(291)

Finally,
H
xH S
P 2
x2H S
H KW
2
=
P KE KS xM xH2 S

yH
=
2S

x2H

(292)

2S

xM xH2 S

with
=

H KW
P KE KS

130

(293)

Distillation

5.1
5.1.1

Introduction
Definitions and aims

Distillation is the process of heating a liquid solution, or a liquid-vapour mixture, to derive off a vapour and then collecting and condensing this vapour.
The products of a distillation process are most often limited to an overhead
distillate and a bottoms, whose compositions differ from that of the feed.

Distillation is one of the oldest and common method for chemical separation.
Historically one of the most known application is the manufacture of spirits
from wine.
Today many industries use distillation for separation within many categories
of products: petroleum refining, petrochemicals, natural gas processing and,
of course, beverages are just some examples.
The purpose is typically the removal of a light component from a mixture
of heavy components, or in the other way around, the separation of a heavy
product from a mixture of light components.

5.1.2

From flash...

Rearranging the figure from the flash cascade at constant pressure gives the
following:
131

Each blue box represents a flash drum, which is a stage of the flash cascade.
Each stream is heated up and cooled down exactly as in the configuration
of the flash cascade. However, when arranged like this, it is easier to note
that, at every stage, there are two streams entering: a liquid which needs to
be heated up, and a vapour which needs to be cooled down.
Instead of providing heat to one stream and subtracting the heat to the
other, it is more convenient to have two streams exchange heat with each
other in a single heat exchanger:

132

But, in every stage, the liquid and the vapour streams dont only exchange
heat, but also matter. Therefore it becomes more convenient to completely
eliminate the heat exchangers and put the two streams in direct contact
with each other.
Moreover, in order to keep the flow rate constant along the column, part of
the two outlets V1 and L5 are partially recycled back to the top and bottom
stage, respectively.
Part of the vapour V1 is condensated at the top of the cascade and fed back
to the first stage. Part of the liquid L5 , collected at the bottom, is again
vaporised in a reboiler and fed back to the last stage of the cascade.

133

5.1.3

... to distillation

A feed mixture F of composition zF enters the column at a certain point,


known as feed stage.

Depending on the thermodynamic characteristics, the incoming feed can


partially split in a liquid and a vapour stream which will join the liquid
and vapour streams already flowing into the column. The vapour stream
flows upwards counter-currently with respect to the liquid stream, which
falls down the column. The vapour becomes richer in the more volatile
134

compound and leaves the column at the top, while the liquid becomes richer
in the hight boiling compound and leaves the column at the bottom. All
of the vapour condensates in the total condenser (yellow), and part of it
is withdrawn, forming the product of the distillation, the distillate [D, xD ].
The other part of the condensate vapour is recycled back into the column
as the reflux [L, xD ]. The reflux enters the column at stage 1 and flows
downwards, together with the new feed and the internal liquid stream.
In order to keep the flow-rate constant in the rectification section of the
column, a portion of the distillate, called reflux, is fed back to the head of
the column.
Besides the column and its internals, in a typical distillation process scheme,
the following key components are also required:
A condenser to cool and condense the vapour leaving the top of the
column. Most often a total condenser is used but a partial condenser
can be applied for specific requirements.
A reboiler to provide the heat needed for the distillation process.
The reboiler is normally a partial one, thus meaning that it acts as a
further equilibrium stage in the overall process. A total reboiler can
be also used for specific purposes.
A reflux drum to collected the condensed vapour coming out from
the top of the column and to recycle part of the liquid back to the
column (reflux).
5.1.4

Industrial distillation

Industrial distillation operations are most commonly conducted in trayed


columns. However, packed columns, or combinations of trays and packing
in the same column, can also be found.
Please refer to the contactors section for the description of the column distillation equipment (column and tray types, packing material).
A distillation column, tray or packed, is normally used to carry out a continuous distillation operation.
This refers to the way in which feed and products are introduced to and
withdrawn from the process. In a continuous operation, feed and products
are continuously introduced and withdrawn from equipment and their flow
rates should be equal at any point in time.

135

However, distillation can also occur in batch mode.


This means that the feed is introduced at the start of the distillation process, and the products are discharged at the end of the operation. Once the
equipment is empty, a new load of feed is introduced and distilled.
Typically, batch distillation processes are those within the spirit manufacturing field. A container is filled at the start, then heated. The vapours
forming over time are condensed to obtain the distillate product (alcoholic
drink). When a proper amount of distillate is collected, the operation is
stopped and the container is emptied.
In general, batch distillation is used for small volume production and when
the same container must be used for several different batch processes.
Continuous distillations are used most often when large volume products
are required.
In the further treatment of the distillation topic and for process design calculation, it will be always assumed to deal with continuous unit operations.

5.2
5.2.1

Process description
The column streams

In the following, the different streams in a distillation column will be highlighted.


Feed: The feed is generally a gas/liquid mixture with the feed flow
rate F [mol/h] and the mole fraction of the more volatile component,
zF , containing at least 2 components.

136

Distillate and bottom: The distillate is a saturated liquid of the flow


rate D [mol/h] and the mole fraction of the more volatile component,
xD , which is the main component.
The bottoms is also a saturated liquid, with the flow rate B [mol/h]
and the mole fraction of the more volatile component xB . It mainly
contains the less volatile component.

Column streams: The internal liquid and vapour are the streams
flowing through the plates. Above and below the feed stage, they are
137

different because of the feed entrance.

Reflux: Part of the condensed liquid is recycled back to the column


(L: Flow rate of the recycle [mol/h], xD : mole fraction of the more
volatile component).

138

5.2.2

The distillation column

In the following, the different parts of a distillation column will be highlighted.


In the next figure, the total condenser can be seen. Cooling liquid enters
on the left, flows through a tube bundle, and exits again on the left. The
vapour from the column enters from the top and the condensate leaves from
the bottom.

The following two figures show the rectifying and stripping sections of column, which are situated above and below the feed, respectively. Vapour
moves up and liquid moves down.

The feed stage, as can be seen in the next figure, mostly looks like the other
stages, apart from the inlet.

The partial reboiler in the following figure has heating medium entering
from the left, flowing through a tube bundle, and exiting on the left again.
139

The liquid enters from the bottom, is heated, and leaves the top as vapour.
The liquid can also leave the column completely on the right.

5.2.3

The working principles: thermodynamics

In every stage the liquid and vapour streams come in contact with each
other and exchange heat. This causes on one side the vaporisation of part
of the liquid, and, on the other side, the condensation of some vapour.
The distillation separation works because every time the liquid vaporises, the
more volatile component tends, for thermodynamics reasons, to concentrate
in the vapour. The vapour, richer in the more volatile component, flows upwards to the next stage where a further vaporisation/condensation occurs,
thus becoming more and more concentrated in the more volatile component.
The feed mixture can be then separated into its component fractions because of their significantly different boiling points. Therefore the distillation
can not be successfully applied to separate components with similar boiling
points.
While the pressure is assumed constant over the whole column, the temperature gets continuously higher towards the bottom (giving the highest
temperature in the reboiler and the lowest temperature in the distillate).

140

Moving up from stage 3 to stage 2 of the column, the vapour gets richer
in the more volatile component (y2 > y3 ), while the liquid flowing downwards gets poorer in more volatile component (x2 > x3 ). This is because the
equilibrium compositions in the stage depend on the temperature (T2 < T3 ).

5.3
5.3.1

Binary distillation design


Problem definition

The feed mixture can contain only two components. In this case we are dealing with a binary distillation. Binary distillation problems are a lot simpler
to solve and straightforward graphical methods can be applied. Therefore
our theoretical examination will start with looking at binary distillation
problems.
However it must be said that most of the commercial distillations are multicomponent (the feed is a mixture of chemicals). A good example is petroleum
refining. Crude oil is a very complex mixture of hydrocarbons with thousands of different molecules present. Multicomponent distillation are theoretically and in practice significantly more difficult to calculate and solve.
For binary distillation, we have:
Data: either:
1. q (quality of the feed)
2. TF
141

3. F , zF , hF
Specifications: either:
1. TD
2. xD , xB , R, f , hD
Unknowns: D, B, QC (heat at condenser), QR , L, V (internal flow
rates in rectification section), L0 , V 0 (internal flow rates in the stripping
section), N , f (position of feed stage)
The data on the TRef lux is needed to know whether the condenser is total or
partial or, in other terms, whether the distillate product is a saturated liquid.
We are assuming that the column operates at constant pressure P , the same
pressure of the feed, and with zero heat leak.
hD and hB can be read in the enthalpy-composition diagram of the system
(e.g. like this for the system water-ethanol), on the saturated liquid curve.
Sometimes instead of the specification on the products compositions, the
fractional recoveries of the Distillate F RD = (DxD )/(F zF ) and of the Bottoms F RB = (BxB )/(F zF ) can be given instead.
With the constant pressure, zero leak and CMO (constant molar overflow)
assumptions, a degree of freedom analysis around the column gives C + 6
degrees of freedom, with C being the number of components.
The CMO assumption states:
The internal liquid and vapour flow rates of the rectifying section are
assumed constant. They are indicated with L and V .
The internal liquid and vapour flow rates of the stripping section are
assumed constant. They are indicated with L0 and V 0 .
It is assumed that ideal equilibrium conditions are reached each stage.
In case of binary mixture (C = 2), the analysis gives 8 degrees of freedom.
Therefore, when 8 variables are set, the design problem is completely specified and all the unknowns are unequivocally calculated.
Generally the data about the feed is known (C+2) and additional 4 specifications on the products must be given, with a total of 8 for binary distillation.

142

5.3.2

External balances: mass balance around the column

Regarding the mass balance around the blue area:

we can say, that


F =D+B
F zF = DxD + BxB

(294)
(295)

regarding the more volatile component. From the external mass balance, it
is possible to calculate D and B.
5.3.3

Constant Molar Overflow (CMO) assumption

Rectifying section

143

Figure 1: Please replace indices n with indices j.


For every plate j = 1, .., f 1 the following equations can be written:
1. Overall mass balance:
Vj+1 = Lj + D

(296)

2. Mass balance on the more volatile component:


Vj+1 yj+1 = Lj xj + DxD

(297)

3. Overall heat balance


Vj+1 Hj+1 + QC = Lj Hj + DhD

(298)

4. Equilibrium correlation
yj = f (xj )

(299)

This is a system of 4 equations and 4 unknowns.


Starting from j = 1, one can use the equilibrium correlation to get x1
and the others to find L1 , V2 and y2 .
This procedure can be repeated until j = f 1 (f is the feed stage)
to calculate all the internal streams of the rectification section.
144

Stripping section

V with L0 , V 0 .
Figure 2: Please replace indices m with indices k and L,
For every plate k = f + 1, .., n the following equations can be written:
1. Overall mass balance:
L0k1 = Vk0 + B

(300)

2. Mass balance on the more volatile component:


L0k1 xk1 = Vk0 yk + BxB

(301)

L0k1 hk1 + Qr = Vk0 Hk + BhB

(302)

3. Overall heat balance

4. Equilibrium correlation
yk1 = f (xk1 )

(303)

The same procedure as in the the rectification section can be applied here
at the stripping section to get all the internal streams.

145

The CMO approach assumes that the amount of molecules which evaporate
and which condensate are the same. This means that within each section
the liquid and the vapor flowrates remain constant in the whole section.
This can be translated into the following equations:
Rectification section:
L0 = L1 = . . . = Lj1 = Lj = . . . = Lf 1 = L

(304)

V1 = V2 = . . . = Vj1 = Vj = . . . = Vf 1 = V

(305)

Stripping section:
L0f = L0f 1 = . . . = L0k1 = L0k = . . . = L0N = L0
Vf0

Vf01

= ... =

0
Vk1

Vk0

= ... =

VN0

=V

(306)
(307)

The CMO assumption implies also that the heat of vaporisation () of the
two components of the feeding mixture must be the same.
It can be verified that this is a reasonable assumption (e.g. for the system
EtOH-Water it is: EtOH = 9.2 [Kcal/mol] and Water = 9.7 [Kcal/mol].
5.3.4

Operating lines

Rectification section: Lets consider the red envelope in the upper


part of the column:

1. Overall mass balance:


V =L+D

146

(308)

2. Mass balance on the more volatile component:


V yj+1 = Lxj + DxD

(309)

3. Equilibrium correlation:
yj = f (xj )

(310)

Substituting the first into the second equation and solving with respect
to yj+1 yields:
L
L
(311)
yj+1 = xj + (1 )xD
V
V
Eliminating the generic index j from the previous equation, we obtain
the generic equation of the operating line for the rectification section:
y=

L
L
x + (1 )xD
V
V

(312)

In the composition diagram, the operating line is a straight line passing


through the point xD y1 and with slope = VL .

From the general mass balance, one can prove that since L/V = 1
D/V , the slope L/V is positive and 0 < L/V < 1.
Stripping section: Lets consider the blue envelope in the lower part
of the column:

147

For the section contained in the blue envelope, the following equations
can be written:
1. Overall mass balance:
L0 = V 0 + B

(313)

2. Mass balance on the more volatile component:


L0 xk1 = V 0 yk + BxB

(314)

3. Equilibrium correlation:
yk = f (xk )

(315)

Again, first equation into second, solve for yk :


yk =

L0
L0
x

(
1)xB
k1
V0
V0

(316)

As well as for the rectification section, we eliminate the generic index


k and we obtain the generic equation of the operating line for the
stripping section:
L0
L0
y = 0 x ( 0 1)xB
(317)
V
V
In the composition diagram, the operating line is a straight line passing
0
through the point xB = xN +1 and with slope = VL 0 .

148

From the overall mass balance, one can prove that since L0 V 0 = B > 0,
that L0 > V 0 and the slope L0 /V 0 > 1.
5.3.5

Number of stages: McCabe Thiele procedure

The McCabe-Thile approach is a graphical method which allows for the calculation of the number of stages of a column in the binary distillation case.

The method is applied to both the sections of the column and it is based to
the following observations:
xD is a data of the distillation design hence it is known. y1 xD holds
for the hypothesis of total condenser.

149

The liquid and vapour streams leaving a stage are in equilibrium.


This means that e.g. for stage 1, x1 and y1 are linked with each
other by the equilibrium correlation and the compositions can be read
graphically on the equilibrium curve in the x-y diagram.
The compositions of the two counter-current liquid and vapour streams
are represented by the operating line equation, e.g. for stage 1, they
are x1 and y2 . These composition can be read graphically on the
operating line represented in the x-y diagram.
y=

L
L
x + (1 )xD
V
V

(318)

Rectification section: The operating line equation above can be


used together with the equilibrium curve for the graphical calculation
of the number of stages.
The starting point is (xD , y1 ), since this is a known value.
The first step is to find the composition x1 in equilibrium with y1 .
This is simply the abscissa of y1 .
(x1 , y1 ) are the compositions of the streams leaving the first stage of
the column.
The point x1 is correlated with y2 through the operating line equation:
y2 =

L
L
x1 + (1 )xD
V
V

(319)

This means that it is on the operating line, ordinate of x1 .


In the same way, the following stage 2 (x2 , y2 ) can be found on the
equilibrium curve, and so on and so forth for the other stages.
The steps keep getting smaller and smaller, while approaching the
point P with a theoretically infinite number of stages.

150

Stripping section: The same observations of above are valid for the
stripping section and the McCabe-Thiele graphical constructionis also
possible:
The operating line equation
y=

L0
L0
x

(
1)xB
V0
V0

(320)

can be used together with the equilibrium curve for the graphical calculation of the number of stages.
The starting point (xB , yn+1 ) is part of the data and a point of the
operating line (from the mass balance around the reboiler).
The composition yn of the vapour in equilibrium with the Bottoms
composition xB is found on the vertical line from xB .
(xB , yn ) are the compositions of the streams leaving the partial reboiler (note: the reboiler is partial, therefore stage n is the reboiler).
The composition yn is correlated with xn1 through the operating line
equation
L
L
yn = xn1 + ( 1)xB
(321)
V
V
which means that it is on the operating line, as an abscissa of yn .

151

In the same way, the following compositions of the stage n1 (xn1 , yn1 )
are on the equilibrium curve (and so on...). Again, we approach the
point P in infinite stages.

5.3.6

Feed quality

q is called quality of the feed, and is defined as


H hF
L0 L
=
=q
F
H h

V0V
=q1
F

(322)

These expressions come from mass and heat balances around the feed stage:

1. Mass balance:
F + V 0 + L = V + L0

(323)

F hf + V 0 H + Lh = V H + L0 h

(324)

2. Heat balance

(the CMO allows us to assume: HV = HV 0 and hL = hL0 ).


q is typically a design specification and allows us to identify the status of
the feed (vapour, liquid or mixture) :

152

When q = 1 hf = h F is a saturated liquid.


When q = 0 hf = H F is a saturated vapour.
When q > 1 F is a subcooled liquid.
When q < 0 F is a supersaturated vapour.
More in general: 0 < q < 1 the feed is a liquid/vapour mixture.
A very interesting application of the parameter q is the so-called q-line.

The q-line is an equation representing the locus of all the intersection points
of the rectification and the stripping operating lines. The derivation is really
simple and comes from the equations we wrote for the derivations of the
operating lines, which we can rewrite in a more convenient way:
V y = Lx + DxD
0

V y = L x BxB

(325)
(326)

Subtracting these two equations term by term gives us


y=

L0 L
F
x 0
zF
0
V V
V V

(327)

Inserting the definition of the quality of feed, q, we get the most convenient
equation for the q-line:
y=

q
1
x
z
q1
q1

153

(328)

5.3.7

The reflux ratio

The reflux ratio indicates the amount of product to be recycled back to the
top of the column. The recycle is strictly needed to keep the internal flow
rates constant inside the column. The reflux ratio is defined as follows:
R=

L0
L
=
D
D

(329)

(the CMO assumption lets us assume L0 = L).


The reflux ratio is linked to the slope of the operating line of the rectification
section in the following way:
L
R
=
V
R+1
R=

L
V

L
V

(330)
(331)

These two equations come from the mass balance around the condenser,
V = L + D.

Dividing both the terms by D and rearranging, one gets the following steps:
V =L+D
V
L
=
+1
D
D
V
=R+1
V L
L
1
1
=
V
R+1
L
R
=
V
R+1
154

(332)
(333)
(334)
(335)
(336)

The reflux ratio is one of the specifications for the column design, and consequently also the operating line for the rectification section is defined and
can be drawn in the composition diagram.
The two limiting values of the reflux ratio are Rmin and R .
The reflux ratio is used in column design in the following way:
The q-line can be represented in the composition diagram. It is a straight
line of the equation
q
1
y=
x
zF
(337)
q1
q1
It intersects the diagonal at the composition zF .
Usually, the quality of feed q is given, so the q-line is easily drawn. It represents the intersection points between the operating lines of the rectification
and stripping sections. Depending on the slope of the operating line of the
rectification section VL , the point of intersection changes. At a certain slope,
( VL )min , the point of intersection lies on the equilibrium line and is called
pinch point P . As we already know, we need infinite stages to reach it. A
distillation designed like this cannot take place. Once the q- line is drawn,
the pinch point is unequivocally determined, and therefore also ( VL )min . In
order to carry out the required distillation, it must hold that:
 
L
L
>
(338)
V
V min

Rmin can be given as a specification of the design or can be calculated


graphically or analytically thanks to the relationship with the operating
155

line of the rectification section:


 
L
Rmin
=
V min Rmin + 1

L
Rmin =

min

L
V min

(339)
(340)

Rmin defines the minimum amount of recycle, at which the distillation task
can be carried out with an infinite number of stages.
When R , the slope of the rectification section operating line coincides with the diagonal (L/V = 1) and the number of stages is the minimum
to perform the required distillation task.
The final reflux ratio is calculated by simply increasing the value of Rmin
with a multiplying factor:
R = Rmin (1.05 2)

(341)

Of course the choice of the right multiplying factor to use is based on economical considerations, like running cost vs. investment of capital, which
are specific to every plant.
5.3.8

Heat balances

Overall balance around the column:


F hF + Qc + Qr = DhD + BhB

(342)

The heat is assumed to be positive when provided to the system (Qr )


and negative when removed (Qc ).
From the external mass balance, we saw how to calculate D and B.
To calculate Qc and Qr , the inside of the column has to be examined,
with particular notice of the total condenser.
Balances around the total condenser:
1. Overall mass balance:
V1 = L0 + D = D(1 + R)

(343)

2. More volatile component mass balance:


V1 y1 = D(1 + R)xD

(344)

V1 H1 + Qc = (L0 + D)hD

(345)

3. Overall heat balance:

156

We know R, H1 (from the enthalpy diagram, since V1 is a saturated


vapour) and we can calculate V1 .
From the heat balance, it is now possible to calculate Qc and finally
plug it into the overall heat balance around the column to find Qr .
5.3.9

Final design

Step by step procedure to design a distillation column:


Step A: Set up the external balances to calculate D and B.
Step B: Apply the McCabe-Thiele graphical construction to calculate N and
f.
Step C: Set up the heat balances to calculate Qc and Qr .

5.4
5.4.1

Multicomponent distillation design


Problem definition

For feeding mixtures containing C species (multicomponent), the C + 6 degrees of freedom are saturated with the additional C 1 compositions given
for the feed.
The unknowns remain the same as for the binary case while the specifications can not be given for all of the C components but only for two, called
key components, for which the distillation is designed. The key components
are the two most important. The aim of the distillation is therefore to
separate these two components from each other.
Of these two chosen key components, the more volatile is called light key
(LK) and will be found mostly on the top product. The higher boiling is
called heavy key (HK) component and will be separated at the bottom of
the column.
xD,LK , xB,HK are then the specifications, i.e. the required compositions of
these two components respectively in the top and in the bottom products
of the distillation.
Concerning the other C 2 components, they are called light not key (LN K)
when they are more volatile than the light key and heavy not key (HN K)
when they are less volatile than the heavy key.
Here a recap of the starting data, specifications and unknowns for a multicomponent design task:
157

The following unknowns must be then calculated:

5.4.2

The external balances issue

Let us just focus only on external mass balances as we already did for the
binary distillation case. The heat balances are decoupled from the mass
balances, therefore they will be not taken into consideration in this section.
Let us write down the external mass balances and the given specifications:
158

While the unknowns are the following:

It is easy to observe that:

In the multicomponent distillation case, C 2 equations and/or specifications are still missing in order to solve the external balances. For this reason,
in order to completely specify the system, additional C 2 assumptions must
be made.
5.4.3

(C 2) assumptions to solve the external balances

As an example, let us consider a feed of 4 components 1, 2, 3 and 4 to be


separated by a distillation process. The volatility increases in the direction
of the component 1:
1 > 2 > 3 > 4
(346)
The aim is to separate component 2 from component 3 with a distillation
process. Therefore these are chosen as the key components. It follows:
159

Component 1: Light not key (LNK)


Component 2: Light key (LK)
Component 3: Heavy key (HK)
Component 4: Heavy not key (HNK)
The specifications on these two components must be given. For instance:
Component 2: xD,LK = 0.9
Component 2: xB,HK = 0.9
while for the other two components the following assumption can be made:
Component 1: xD,LN K = 1 which is the same as xB,LN K = 0
Component 4: xB,HN K = 1 which is the same as xD,HN K = 0
or in other words:
All of component 1 is all present in the Distillate stream. This is reasonable since it is more volatile than component 2, which is separated
mostly at the top.
All of component 4 is all present in the Bottoms. This is also reasonable since it is less volatile than component 3, which it is separated
mostly at the Bottoms.
In case of C components, this assumption is true for all the LNK and HNK
components, of which there then are C 2.
In order to easily solve the external mass balance, instead of the mole fraction, it is more convenient to use the fractional recovery with respect to the
Distillate and to the Bottoms, as defined here:

F Ri,D =

DxD,i
F zi

F Ri,B =

BxB,i
F zi

(347)

The assumptions can be then rewritten in terms of fractional recovery as


follows:
Component 1: F RD,LN K = 1 or also F RB,LN K = 0
Component 4: F RB,HN K = 1 or also F RD,HN K = 0
while the specifications can be rewritten as follows:
160

Component 2: F RD,LK = 0.9


Component 3: F RB,HK = 0.9
In order to calculate the flow rates D and B, as well as the top and bottom
compositions, the definitions of the fractional recovery above leads to the
following equations:
X

DxD,i =

i=1..c

F RD,i F zi

i=1..c

BxB,i =

i=1..c

F RB,i F zi

(348)

F RB,i zi

(349)

i=1..c

D and B can be then calculated as follows:


D=F

F RD,i zi

B=F

i=1..c

X
i=1..c

from which finally also the top and bottom compositions are calculated:

xi,D = F zi
5.4.4

F RD,i
D

xi,B = F zi

F RB,i
B

(350)

Approximate shortcut methods

Although rigorous calculation techniques are available, it is common practice to use the so called approximate methods in order to get preliminary
design and/or to optimize the design conditions for a multicomponent distillation problem.
The method, we are going to illustrate in this section is known under the
name of Fenske-Underwood-Gilliland (FUG), after the three guys which
developed the three different parts of the method in order to get the ideal
number of stages of a multicomponent distillation column.
The method is composed of the following steps:

161

5.4.5

Fenske equation

The Fenske equation allows for the calculation of the minimum number
of stages. The final equation is shown below in two forms, to be used
respectively whether composition or fractional recovery specifications of the
key components in the Distillate and in the Bottoms products are given.
(x

Nmin =

/x

LK

HK B

(351)

ln LK,HK
(F R

Nmin =

ln (xLK /xHK )D

) (F R

HK B
ln (1(F R LK) D)(1(F
R
LK D

ln LK,HK

) )

HK B

(352)

where the constant of relative volatility LK,HK is calculated with respect to


the heavy key component, taken as a reference:
LK,HK =

kLK
kHK

(353)

LK,HK depends on the temperature, therefore it is different at every stage


of the column.
However, most often, for the sake of simplicity, it is possible to use a constant
value valid for all the column stages. When this is not possible, an average
value of LK,HK among all the (N = Nmin ) values must be then calculated,
e.g. average between the value at the first stage and at the reboiler:
average =

1 R

(354)

For the derivation, the concept behind the Fenske equation is very simple.

162

The column depicted here with the hypothesis of total reflux (no feed no
products) is taken into consideration. Then the equilibrium and the mass
balance equation are written for the two key components starting from the
reboiler stage (see envelope in the figure).
The method is applied to the light key and heavy key components, under
conditions of total reflux.
Step 1
Equilibrium equations at the reboiler (stage R):
(yLK )R = kLK (xLK )R

(355)

(yHK )R = kHK (xHK )R

(356)

Dividing (355) and (356) term by term, yields


x 
y 
LK
= R LK
yHK R
xHK R
where
=

163

kLK
kHK

(357)

(358)

Step 2
Mass balance around the reboiler.
Overall mass balance:
V 0 = L0

(359)

Mass balances of the key components


V 0 (yLK )R = L0 (xLK )N
0

(360)

V (yHK )R = L (xHK )N

(361)

Substituting equation (359) in equations (360) and (361), yields:


(yLK )R = (xLK )N

(362)

(yHK )R = (xHK )N

(363)

Step 3
The liquid composition of the reboiler stage, which is a given specification, is correlated with the liquid composition of the previous stage
N, which can be consequently calculated.
Dividing equations (362) and (363) term by term and substituting
equation (357), yields:
x

LK

xHK

= R

x

LK

xHK

(364)

Step 4
This iteration proceeds backwards from stage N, to N-1, to N-2, and
so on so forth until it reaches the liquid composition of the distillate
(stage D), which is known as well.
For the stage N-1, it is:
x

LK

= N R

x

LK

xHK N 1
xHK R
...until stage D for which it can be written:
x 
x 
LK
= 1 2 . . . N 1 N R LK
xHK D
xHK R

(365)

(366)

assuming and average value of valid for all the stages, equation (366)
can be written as:
x 
x 
Nmin
LK
= aver.
LK
(367)
xHK D
xHK R
in equation(367) the only unknown is the minimum number of stages
(including the reboiler), which can easily be derived.
164

5.4.6

Underwood equations

The Underwood equations allow for the calculation of the minimum reflux
Rmin .

1q =D

c
X
i,HK zi
i,HK

Underwood I

(368)

Underwood II

(369)

i=1

Vmin

c
X
i,HK xD,i
=D
i,HK
i=1

where is a particular constant of relative volatility and


LK > > HK

(370)

For the sake of simplicity we will not explain why the value of is fixed in
this way.
It is only important to underline that this value can be so chosen only when
two key components are adjacent to each other in the volatility scale:
LN K > LK > HK > HN K

(371)

Procedure for the calculation of the minimum reflux:


Step 1. Apply the first Underwood equation for the calculation of .
Step 2. Apply the second Underwood equation for the calculation of Rmin :
Rmin =

Lmin
Vmin
=
1
D
D

(372)

The last step is the calculation of the finite reflux. This is done, as for the
binary case, using a multiplying factor:
R = (1.05 2) Rmin

(373)

The Underwood equations are derived as follows.


The method is applied to all i components of a distillation, under the columns condition of minimum reflux. For a binary system, such a condition
is represented in the composition diagram when the operating lines intersect
in the pinch point at stage j. For a multicomponent distillation, the diagram cant be drawn, but the analytical condition for the pinch point can
be written in any case.

165

Step 1
Pinch point condition at stage j: The point representing the concentrations at stage j lies both on the equilibrium curve and the operating
line the equilibrium and the operating line compositions coincide.
xj1 = xj = xj+1

(374)

yj+1 = yj = yj1

(375)

Step 2
Mass balance around the red envelope in the rectification section

Mass balance for all i components:


Vmin yj+1,i = Lmin xj,i + D xD,i

(376)

Step 3
Equilibrium condition for component i at stage (j + 1):
yj+1,i = ki xj+1,i

(377)

and, because of the pinch condition:


yj+1,i = ki xj,i

(378)

where the distribution coefficient of component i can be expressed


through its constant of relative volatility referred to a component r,
which most often is the heavy key component (HK):
ki = i,r kr

166

(379)

Step 4
Substituting (378) and (379) in (376) and rearranging:
P
X
D xD,i
Vmin yj+1,i =
Lmin
1 Vmin
k

(380)

i,r r

which is convenient to write as follows:


X i,r xD,i
Vmin = D
min
i,r VL
min k

(381)

A similar expression can be derived for the stripping section in the


same way:
0 x
X i,r
B,i
0
Vmin = B
(382)
0
L
min
0
i,r
V 0 k0
min

Step 5
Under the assumptions of CMO and of the constant relative volatility
(CRV) constance along the column:
0
i,r = i,r

(383)

it can be proved that


=

Lmin
L0
= 0 min 0 = 0
Vmin kr
Vmin kr

(384)

The term (or 0 ) has the dimension of a constant of relative volatility.


Step 6
Underwood II: Final expression.
Substituting equation (384) in (381), the final expression of the second
Underwood equation is obtained:
X i,r xD,i
(385)
Vmin = D
i,r
Step 7
Underwood I: Final expression.
Subtracting equations (381) and (382) term by term and substituting
the expression of equation (384) yields:
0
Vmin Vmin
=

X i,r (DxD,i + BxB,i )


i,r
167

(386)

and remembering that from a balance around the feed stage, the following equation is valid:
0
Vmin Vmin
= F (1 q)

(387)

and from the overall mass balance


DxD,i + BxB,i = F zF,i

(388)

Substituting (386) and (387) in (388), the final expression of the first
Underwood equation is obtained:
1q =
Calculation of

X i,r zF,i
i,r

(389)

Equation (389) is a function of :


1q =

X i,r zF,i
= f ()
i,r

(390)

whose derivate is always positive, as follows:


X i,r zF,i
df
=
2 > 0
d
i,r

(391)

This function can be represented as follows (blue line):

c solutions are possible. 1 is negative hence meaningless. Among the other


c 1 meaningful solutions, the solution is the value of staying between
the LK and HK , when the key components are adjacent in the scale of
volatility:
LK > > HK

168

(392)

5.4.7

Gilliland correlation

Gilliland used an empirical correlation to calculate the final number of stage


N from the values calculated through the Fenske and Underwood equations
(Nmin , R, Rmin ).
The procedure is very simple and uses a diagram like the one shown here
below.
One enters the diagram with the abscissa value, which is known, and reads
the ordinate of the corresponding point on the Gilliland curve.
The only unknown of the ordinate is the number of stages N .

Other authors, besides Gilliland, have developed similar empirical correlations or have tried to find a mathematical expression for the Gilliland
correlation. Here below some of the most significant results:

169

5.4.8

Position of feed stage

In order to determine the position of the feed stage, it is assumed, that the
composition of the feed is the same as the composition at the feed stage:
xF = xF eedstage

(393)

Now, Fenskes method is used, replacing the compositions of the bottoms


by the composition of the feed.
React.
Nmin

ln
=

(xLK /xHK )D
(xLK /xHK )F

ln LK,HK

(394)

Having already determined the minimum (Nmin , Fenske) and theoretical (N ,


Gilliland) number of stages, the following correlation holds
React.
Nmin
N React.
=
(395)
Nmin
N
This gives the number of stages for the rectification section, and therefore
the position of the feed stage.

5.5
5.5.1

Non-standard distillation
Partial condenser

Lets consider the red envelope around the condenser. In this case the partial
condenser acts as a further flash drum, hence it can be seen as a further
170

equilibrium stage.

The distillate is a vapour stream with composition yD , which is now given


as a specification as it was xD for a total condenser.
Of course yD 6= y1 since a further equilibrium stage takes place in the condenser.
The mass balances around the condenser remain the same:
1. Overall mass balance
V =L+D

(396)

2. Mass balance for the more volatile component


V y1 = Lx0 + DyD

(397)

Consequently the operating line for the rectification section is:


y=

L
L
x + (1 )yD
V
V

(398)

From a graphical point of view, nothing changes for the McCabe-Thiele


graphical construction for the number of stages calculation since the operating line passes through the point x = y = yD , as it was for the case of the
total condenser.
However from a conceptual point of view, the first stage is no physical plate
of the column anymore but represents the partial condenser.

171

5.5.2

Total reboiler

In case of total reboiler, the entire liquid stream coming out of the last stage
of the column is entirely vaporized. The vapour composition is the same as
the liquid composition: xn = yn+1 = yB . yB is now a specification of the
design problem.

Lets consider the blue envelope around the reboiler. The mass balances
remain the same as with the partial reboiler:
1. Overall mass balance:
L0 = V 0 + B

(399)

2. Mass balance for the more volatile component


L0 xn = V 0 yn+1 + ByB
172

(400)

The equation for the operating line also remains the same:
y=

L0
L0
x ( 0 1)yB
0
V
V

(401)

From the graphical point of view, the McCabe-Thiele construction for the
calculation of the number of stages remains the same as for the case of the
partial reboiler, since the following equation is satisfied for x = y = yB .
However from a conceptual point of view, the last stage does not correspond
to the reboiler any more but is a physical plate of the column.

5.5.3

Open steam (or direct steam)

This kind of configuration is used when the feed mixture to separate is a


water based mixture containing a light component.
In a normal column configuration with partial reboiler, the bottom product
would be an almost pure water stream while the light component is separated as top product of the column. This highly water concentrated stream
is sent to the reboiler, thus becoming steam and it is refed into the last stage
of the column.
For the heating up process, the reboiler also uses steam. This steam can be
then directly fed into the column, thus saving money for capital investiment
(to buy the reboiler) and running cost (use of the steam to produce other
steam).
The stripping part of the column can be schematized as follows:

173

If we consider the blue envelope in the bottom part of the column. The
mass balances are the following:
1. Overall mass balance:
V 0 + B = L0 + S

(402)

2. Mass balance for the more volatile component


V 0 yk + BxB = L0 xk1 + Sys

(403)

From the CMO assumption, it also holds, that V 0 = S and L0 = B, giving


the equation for the operating line
y=

L0
(x xB )
V0

(404)

This equation is satisfied for x = xB and y = 0.


In the composition diagram, the graphical McCabe-Thiele construction can
be applied for the calculation of the number of stages.
xB is not on the diagonal anymore as it was with standard column, but on
the abscissa-axis (y = 0).
5.5.4

Side streams

This kind of column configuration is typical of the petrochemical plants,


where the most common running unit operation is the fractional distillation.
It consists of splitting a mixture of various components, the crude oil, into
its components. Because of their different boiling temperatures, the components (or so-called fractions) of the crude oil are separated at different levels
174

(i.e. plates) of the column, where different boiling temperatures are present.
The fractions are then withdrawn from their respective plates, therefore the
column presents numerous side streams.
For the time being, lets simply take into consideration following column
with just one side stream.

The column is now divided in three sections, with one section being a segment of stages between two input or two output streams:

175

As well as the other sections, it is possible to derive an operating line for


the middle section. Because of the withdrawal of one stream, the internal
flow rates of the middle section change. Lets assume they are now L and
V .
The following additional data are introduced to the design problem:
The flow rate S of the side stream
The quality of the side stream: liquid or vapour
The composition ys or xs of the side stream
Writing the mass balance for the the red dotted line, the operating line of
the middle section can be easily derived:
y=

DxD + Sxs
L
x+
V
V

(405)

The operating line can be drawn in the composition diagram in order to


calculate the number of stages with the McCabe-Thiele construction.
To draw the operating line of the middle section, from the new data given
we can:
draw the q-line of the side stream: horizontal if vapour, vertical if
liquid
176

identify the intercept between operating line of the rectification section


and the q-line of the side stream (1st point of the operating line).
identify the intercept of the operating line for the middle section with
the diagonal (2nd point of the operating line):
y=x=

DxD + SxS
D+S

(406)

Given two points, the operating line of the middle section can now easily be
drawn.
The McCabe-Thiele graphical construction for a saturated liquid side stream
is shown.

5.5.5

Multiple feeds

Most of the time, in the industrial plant, a distillation column is just a piece
of a more complex separation unit which involves many other processes in
parallel. Flash, absorption and distillation units are very often coupled together to obtain the final product.
Actually the key objective in a plant is to create the least waste amount
possible. Of course for economical and environmental reasons.
Therefore every kind of exit stream, if not already the final product, is recycled as raw material for other productions or it is further treated and
separated to obtain different final products.
For this reason, a distillation column can very often present more than one
feed streams. Since the feed streams are already partially separated, they

177

are fed into the column at different plates, in order to maintain the separation that already exists.
Lets take into consideration the following column with two feed streams.

The column is now divided into three sections, a section being a segment of
stages between two input and two output streams.

178

Like for the other sections, it is possible to derive an operating line for the
middle section. Because of the feeding of one stream, the internal flow rates
of the middle section change. Lets assume they are now L and V .
The following additional data is introduced in the design problem:
the new feed F2
the quality q2 of the feed F2
the composition zF2 of the new feed.
In order to derive the operating line for the middle section, we write a mass
balance for the part of the column delimited from the black dotted line at
the top of the column (see figure above).
Overall:
D + L = V + F2

(407)

For the more volatile component:


DxD + L x = V y + F2 zF2

(408)

We get the operating line:


y=

L
DxD F2 zF2
x+

V
V
179

(409)

The operating line can be drawn in the composition diagram in order to


calculate the number of stages with the McCabe-Thiele construction.
In order to draw the operating line of the middle section, from the new data
given we can:
draw the q-line for the second feed F2 .
identify the intercept between operating line of the rectification section
and the q-line of the second feed (1st point of the operating line).
identify the intercept of the operating line of the middle section with
the diagonal (2nd point of the operating line).
From the mass balances around the top section of the column (see dotted
line in the figure above) it follows that:
y=x=

DxD F2 zF2
D F2

(410)

Given these two points, the operating line for the middle section can be
drawn.
The following figure shows the McCabe-Thiele graphical construction for a
column with two feed streams.

Up to now we have always assumed that in every plate of the distillation


column, the equilibrium conditions are reached. In reality this is almost
never true and the actual plate can be compared to the equilibrium plate
using a concept of efficiency.
At first sight, it seems to be easier to introduce the concept of overall efficiency E0 as simply the ratio between the theoretical number of equilibrium
stages to the actual number of stages required:
180

E0 =

Nequil.
Nactual

(411)

The overall efficiency considers all the phenomena affecting the efficiency in
the column, from e.g. hydrodynamics, like viscosity and flow rate, to the
mass transfer, like diffusion. The difficulty is how to take all these variables
into account to quantify the overall efficiency.
Moreover, this efficiency can refer to the columns plates and doesnt include
the additional number of stages given by the presence of a partial condenser
and partial reboiler. This is because the efficiency of the reboiler or of the
condenser must take into account other and/or different factors.
In other words, the overall efficiency has the advantage that it is easy to use
but the disadvantage that it is very difficult to calculate from principles.
It is rather convenient to estimate a stage efficiency from principles (see also
absorption and stripping efficiency).
The most common used stage efficiencies are those defined by Murphree.
The Murphree efficiency for the vapour side, EM V , is defined as follows:
EM V =

yk yk+1
yk yk+1

(412)

In an (x-y) composition diagram, it can be easily observed that the denominator represents the vertical distance between the operating line and the
equilibrium curve, while the numerator corresponds to the vertical distance
between the operating line and the actual outlet concentration from the
stage (see fig. here below).
The points of the actual concentrations can be connected to form the pseudoequilibrium curve (see figure below).

181

A Murphree efficiency for liquid side, EM L , can also be defined:


EM L =

xj xj1
xj xj1

(413)

The efficiency has an impact on the number of stages. The graphical construction of McCabe-Thiele can be used. The steps are built using not the
equilibrium curve any more, but the new pseudo-equilibrium curve.
The number of stages increases when the efficiency gets lower.
It is always more convenient to use the Murphree efficiency EM L with the
Distillate and to use the Murphree efficiency EM V when dealing with the
Bottoms.
As we saw, the Murphree efficiencies refer to a specific plate (j, k), and are
therefore different for every different stage of the column.
However for the sake of simplicity, here, the same value of efficiency will be
considered for all the plates.
Finally, for the efficiency of the partial reboiler it is assumed: Ereboiler 1.

182

5.6
5.6.1

Azeotropic column
Principles

If a liquid mixture forms an azeotrope, the separation that can be achieved


is normally limited.
As an example we can consider the system Ethanol-Water. This forms a
homogeneous minimum boiling azeotrope at a concentration of 0.8943 mole
fraction ethanol, as shown in the two diagrams here below.

Ordinary distillation of such a system always gives the azeotropic mixture


as distillate, the azeotrope being the minimum boiling compound.
183

In order to pass the azeotropic point, or, as it is usual to say, to break the
azeotrope, ordinary distillation columns can be coupled with other separation methods like adsorption, membranes, extraction and so forth.
In the next figure, a distillation column is coupled with another or a group
of other separators. The incompletely separated stream coming from the
separators is recycled back to the distillation column as a second feed.

Even though this configuration is used in industrial plants, the question always remains when it makes sense to still use a distillation when another
separation method is available to completely separate the system. Of course
also in this case, an economical evaluation must be made in order to choose
the optimum configuration.
Another group of methods consists of modifying the system in such a way
to break the azeotrope and to still use a distillation process to separate
the components. Extractive distillation, azeotrope distillation and pressure
swing distillation are three of the most commom methods, to which the
following sections are dedicated.
5.6.2

Extractive distillation

The idea is to add a solvent which could modify the relative volatility between the two components of the azeotropic system and could enlarge the
gap in between the boiling points of the two components.
184

For this reason, the solvent must be selected so that one of the components
is selectively attracted to it.
Most often, the solvent has a very high boiling temperature with more affinity to the high boiling component of the mixture (B in the next figure).
This results in dramatically increasing the relative volatility of A compared
to the mixture B plus the solvent, separating component A as a distillate of
an ordinary distillation process.
The mixture of component B and the solvent is then further separated with
a normal distillation process.
A typical flow sheet is shown below:

In the flow sheet, two distillation columns are present. Distillation column
2 is only an ordinary one to separate the solvent-B component mixture.
In column 1, the solvent is added in the upper part of the column, a few
stages below the top stage.
The solvent, being very high boiling, falls downward and attracts the B component. In the middle of the column, where the concentration of the solvent
is very high, the complete separation between A and solvent-B is carried out.
At the top, the component A can easily be separated because of the increased relative volatility compared to the solvent-B component mixture.
185

The selection of the solvent is the key point for the entire process.
The solvent must not form an azeotrope with any of the components, in
order to be easily recuperated and recycled. Then, as usual, it is required
that it is nontoxic, nonflamable, nonreactive, available and inexpensive.
Most of the time it is really difficult to find all these characteristics at once.
For this reason, even if its not the best solvent, one good procedure is to
select a solvent from those already used in the plant or which is a byproduct
of other productions.
5.6.3

Azeotropic distillation

Azeotropic distillation is a method of modifying the equilibrium of these


complex mixtures to separate their components.
Usually, two kind of azeotropic distillation are distinguished:
1. Binary systems which form a heterogeneous azeotrope.
2. Binary systems which form a homogeneous azeotrope. In this case, an
entrainer or solvent is added in order to form an azeotrope with one
or both of the components. The system then becomes ternary.
Azeotropic distillation systems are of course extremely more complex, hence
calculations are not as straightforward as for simple distillation columns.
One important aspect which is not negligible any more is the very nonideal equilibrium behaviour of the components and the presence of multiple
phases inside the column (at least 3 phases: 2 immiscible liquids and one
vapour).
The following figure shows the x-y diagram for the mixture of water (component A) and n-butanol (component B), which forms an heterogeneous
azeotrope. At the azeotropic composition, the homogeneous vapour mixture condenses into a liquid system with two different phases of composition
and in equilibrium with each other.

186

In this case the following configuration can be used to separate the two
components A-B:

Three different unit operations are present:


1. A stripping column: Column 1
2. A liquid-liquid settler to separate the two liquid phases and .
3. A distillation column: Column 2
The stripping column receives the liquid phase of composition x from the
settler. In column 1 the phase is more volatile than the water (component
A), which is then found almost pure at the bottom of this column.

187

The vapour phase of composition x is then collected at the top of column


1 and sent into the liquid-liquid settler, from where it is recycled back into
the same column 1.
In the column 2 the liquid phase of composition x constitutes the reflux of
the distillation column 2. In this case, the phase is less volatile than the
n-butanol and is found at the bottom of this column 2.
The vapour coming from column 2 is sent into the settler where it is split
again into the two liquid phases. In this case the settler is the key part of
the equipment since it allows to break the azeotrope.

Azeotropic distillation with added solvent When a homogeneous


azeotrope is formed, one method is to add an entrainer (or solvent) which
forms either a heterogeneous azeotrope with one or both the components,
or another homogeneous azeotrope which can be broken with alternative
methods (i.e. extractive distillation, pressure swing distillation,etc...).
In the case of a heterogeneous azeotrope, the same procedure as explained in
the first paragraph can be used. Since three components are present, more
than one binary azeotrope or also a ternary azeotrope can be formed.
The mixture ethanol-water can be broken using pentane as entrainer. This
produces the formation of a heterogeneous ternary azeotrope. The next
figure shows an industrial scheme for separation:

Selecting a solvent for azeotropic distillation is often more difficult than


for extractive distillation. There are usually fewer solvents that will form
azeotropes that boil at a low enough temperatures to be easily removed in
the distillate or boil at high enough temperature to be easily removed from
the bottoms.
188

Additionally, the binary or ternary azeotrope formed must be separated.


In practice this requirement is met by heterogeneous azeotropes and by
azeotropes that are easily separated with water wash.
Other requirements for the solvent can be found in the extractive distillation
section.
5.6.4

Pressure swing distillation

The ethanol-water system has an azeotrope at 0.8943 mole fraction ethanol


at 1 atm pressure [see also principles of this section].
If the pressure is reduced from 760 to 100 torr, it is observed (see next
figure), that the composition of the azeotrope goes more and more towards
1 before completely disappearing at pressures below 70mmHg [not visible in
the figure].

At this pressure, the separation could be done in an ordinary distillation


column. Unfortunately the costs to achieve disappearance of the azeotrope
are normally very high and the separation would, in any case, require a very
high and large distillation column. Therefore a distillation process operating
at these conditions is not convenient.
However even if it doesnt completely disappear, the azeotrope composition
is still affected by the pressure.

189

If the shift in composition (D1 D2 ) due to the change of the pressure


(P1 P2 ) is large enough, one can imagine a process where two distillation
columns operate at the two different pressures P1 and P2 .
Typically, P1 is atmospheric pressure, while P2 is a higher pressure.
The following figure shows the pressure configuration for a system as described above:

The principle is, that the at the first pressure, one distils close to the
azeotrope composition. Switching to a second, usually higher, pressure
pulls the azeotropic composition down. Since the composition leaving the top of the first column is now higher than the composition of the
new azeotrope, the distillation can continue.

190

Liquid-liquid extraction

6.1
6.1.1

Introduction
Definitions and aims

The liquid-liquid extraction or simply liquid extraction is a process where


one or more solutes are removed from a liquid phase, called diluent, and
transferred into a second liquid phase, called solvent, by simple contact of
the two liquids.
The two liquids must be either immiscible or partially miscible so that only
the solute, which has a greater solubility in the solvent phase than in the
diluent, separates from this and moves into the solvent.

Extraction is a common laboratory and industrial unit operation and since it


does not involve any evaporation process, can be carried out also at low temperature, thus making this process so convenient also for high temperaturesensitive products
In the following sections we will focus on the case of partial miscibility between diluent and solvent, hence on extraction with ternary systems.
In the case of total immiscibility, the extraction problem becomes similar
to the absorption and stripping type of problem. In this case the McCabeThiele graphical method, developed for absorption and stripping, can be
easily applied also to design extraction processes.
6.1.2

Process description

The purpose is to recover a certain substance (the solute) at high purity by


the use of a highly selective solvent.

191

In the extraction unit, a solute (yellow) is separated from the feed (orange)
and moves into the solvent (blue). In the further step, in the solvent recovery unit, the solute is finally separated from the solvent, e.g. by distillation
or by other unit operations, and obtained highly pure.

A liquid mixture containing the carrier solvent, called diluent, and the solute is fed to the mixer. The solvent is generally pure and highly selective
for the solute (high solubility).
The feed and solvent phases come into contact and the mass transfer takes
place. During this phase (extraction), the solute moves from the feed phase
to the solvent. We assume that the ideal equilibrium conditions are reached.
The settler allows for the mechanical separation of the two phases due to
their difference in density, which is the driving force for the separation. This
is a critical step since the density difference is never very large (both are
liquids).
The raffinate is the product of the extraction rich in the diluent phase. It
also contains traces of solute and solvent, since these are partially miscible.
The extract is the product rich in solvent and solute. This must be further
treated (e.g. distillation) to recover the solute.

192

6.2
6.2.1

Single stage operations


Single stage: problem definition

The most simple extraction process can be carried out with a simple mixer
and a settler. This is called single equilibrium stage since the feed and
the solvent come in contact in only one mixer (=stage). Ideal equilibrium
condition is assumed.
The problem we face in liquid-liquid extraction can be shown in a ternary
diagram.

The feed is usually a binary mixture of diluent (D) and solute (S). Its properties F , xFA and xFD are usually given DATA. The feed can usually be found
somewhere on the line AD.
The solvent can contain trace amounts of solute or raffinate, but is usually
considered pure. Its composition is therefore always a given DATA. The flow
rate S, on the other hand, is only given DATA in a verification problem,
while being an UNKNOWN in a design problem. The solvent can be found
in the far left bottom corner of the diagram (when it is considered pure,
which it is most of the time).

The mixer is assumed an equilibrium stage, the feed and the solvent ex193

change solute until the equilibrium is reached. The settler only separates
the phases, no more mass transfer or thermodynamic phenomena take place.
The final content of solute in the raffinate is given as a SPECIFICATION
of the extraction task. It comes out of an equilibrium stage, therefore its
composition is on the solubility envelope (curved line) in the ternary diagram (ideally far away from point A). The flow rate as well as the content
of solvent and diluent are UNKNOWN.
The extract composition and flow rate E are also UNKNOWNS of the extraction task. The composition must be a point on the solubility envelope
(best close to A) and correlated with the raffinate composition with a tie-line
(equilibrium condition).
6.2.2

Single stage: process design

In the case of a simple mixer (one equilibrium stage), the design consists in
sizing the flow rate of the solvent which must be used to obtain the required
extract composition and in calculating the obtained extract and raffinate
flow rates.
In order to do this, typically a graphical method is used as shown in the
following procedure.
Please, note that for the seek of simplicity the solvent is considered to be
pure. However, in reality, most likely the solvent comes from a regeneration
step and therefore contains still trace of solute.
Step 1. Construct a tie-line through the raffinate composition (SPECIFICATION, xR
A ). This gives the extract composition at the other end of
E
E
the tie-line on the solubility envelope (xE
A , xD , xS ). The raffinate and the
extract are in equilibrium since the exit an equilibrium stage.
Step 2.

Using the overall mass balance


F +S =E+R=M

(414)

and applying the lever-arm rule, the mixing point M is the intersection
between ER and F S.
Step 3. From the mixing point M , the lever-arm rule is applied to determine the flow rates S, R and E.

194

6.2.3

Single stage: solvent range

Decreasing the amount of solvent moves the point M further away from S
and towards F . This means, that a higher tie-line is now chosen, which
would increase the amount of of solute in the raffinate. If R were a specification, the extraction task could not be reached any more.
Increasing the amount of solvent moves the point M closer towards S, again
along the same line. Due to the lower tie-line, the raffinate will now
contain less solute. The maximum amount of solvent corresponds to a mixing point M on the solubility envelope. The raffinate is the purest possible
(min. solute content).

6.3
6.3.1

Multistage operations: cross-current


Cross-current: cascade configuration

When more mixers and settlers (= single stage) are arranged together in a
cascade, the separation in a liquid extraction task can be improved a lot.
Purer raffinate and more concentrated extract can be then obtained.

195

One of the possible configuration for the cascade of stages is the cross-current
configuration.

In the case of a cross-current cascade, the feed remains unchanged from the
single case. It enters the cascade in the first of N stages (each stage is composed of a mixer and a settler).
The solvent is split into N equal parts and fed fresh each stage, where it
comes into contact with the raffinate stream of the previous stage.
Ideal equilibrium conditions are assumed within each stage. The extract
and raffinate coming out of each stage are in equilibrium with each other
(e.g. E1 with R2 , E2 with R2 and so forth).
The extract is withdrawn from each stage and is in equilibrium with the
raffinate leaving the same stage.
The raffinate is fed to the next stage. This is a key point in cross-current
configuration: The raffinate and the fresh solvent are not at equilibrium,
hence the mass transfer (of solute from the raffinate to the solvent) can
continue.
6.3.2

Cross-current: process design

The following section will show the six step process design procedure of a
cascade of stages in a cross-current configuration. Again a graphical method
is used.

196

Please, note that for the sake of simplicity the solvent is considered to be
pure. However, in reality, the solvent most likely comes from a regeneration
step and therefore still contains traces of solute.
The problem design illustrated here considers a simplified design where the
amount of solvent fed to every stage is given.
In reality, most of the time, only the total amount of solvent is known and the
amount to be delivered to each stage can be calculated only after determining
the number of stages. Therefore the following iterative procedure must be
implemented:
An initial value of the number of stages is guessed first together with the
corresponding solvent amount per stage Then, the required number of stages
using that amount of solvent is calculated as illustrated below. Subsequently,
the calculated number of stages is compared with the initial guess and the
iteration continues with the new value until convergence.
For the following procedure, we have:
DATA: F, xFA , xFD , S/N, xSS
R
R
SPECIFICATIONS: xR
A , xD , xS
D E
UNKNOWNS: E, xE
A , xD , xS , N, R

Step 1.

Mass balance around stage 1:


F+

S
= E1 + R1 = M1
N

(415)

Mixing point M1 identification (lever-arm rule).


Step 2. Tie-line through the mixing point M1 : Determination of Extract
E1 and Raffinate R1 .
Step 3.

Lever-arm rule: Calculation of Extract E1 and Raffinate R1 :


M1 E1
R1
=
F + S/N
E1 R1

M1 R1
E1
=
R1
E1 M1

(416)

Step 4. Mass balance around stage 2, then draw a tie line through M2 ,
finally determine R2 , E2 (all just like for stage 1).
Step 5. Repeat for N stages until the raffinate RN reaches the specification. This is the minimum number of stages required to carry out the
extraction task.
197

Step 6. Calculate the total extract E and raffinate R flow rates and their
average compositions. For N = 3:
E = E1 + E2 + E3
xE
A =

6.3.3

E
E
E1 xE
A,1 + E2 xA,2 + E3 xA,3

R = R1 + R2 + R3
xR
A =

R
R
R1 xR
A,1 + R2 xA,2 + R3 xA,3

(417)
(418)

Cross-current: effect of the number of stages on separation

In a cross-current layout, the amount of solvent used in every stage is linked


to the number of stages by S/N . Therefore for the same total amount of
solvent used, if a higher number of stages is used, consequently the solvent
flow rates fed at every stage is smaller.
This results in a smaller degree of separation in every stage, but still, when
compared to a single stage unit, in a higher total degree of separation if an
appropriate number of stages is used.
This is clearly a limitation of the cross-current configuration: the degree of
separation can improve by increasing the number of stages until a certain
limit, due to the fact that for every stage added, the jump in concentration
for each stage gets smaller and smaller.
The following example looks at a simple case, in which an extraction task
is carried out with 1, 2 or 3 stages.
The next figure shows the process design for 2 stages, as already illustrated
above. The solvent flow rate fed at every stage is S/2. R2 and E2 represent
the final compositions for the raffinate and extract streams.

198

Increasing the number of stages from 2 to 3, the amount of solvent fed into
every stage of the cascade decreases from S/2 to S/3. The mixing point
moves from M1 to M100 (lever-arm rule) and the degree of separation reached
in every stage is smaller than in the case of 2 stages. However, using a higher
number of stages, the purity (resp. solute) of the raffinate R300 is higher (less
solute) than in the case of 2 stages, R200 .

Decreasing the number of stages from 2 to 1, the amount of solvent fed to


the stage increases from S/2 to S. The mixing point moves from M to M1
(lever-arm rule), and the degree of separation reached in only stage 1 is a
lot higher than in the case with 2 stages. However with 2 stages, the final
concentration of the raffinate R2 is still better than in the case of 1 stage
(R10 ).

199

6.4
6.4.1

Multistage operations: counter-current


Counter-current: column configuration

The best configuration for a cascade of mixers and settlers (= single stage) is
the so-called counter-current configuration. This layout allows for still better separation for the same number of stages compared to a cross-current
configuration.
The cascade of stages is arranged in a column, called extraction column.
Please, refer to the contactors section to learn more about this topic.

In the counter-current configuration, the fresh solvent comes into contact


with the raffinate at stage 1 and flows counter-currently. The raffinate leaving stage 1 (R1 ) is the final raffinate product.
Again, ideal equilibrium conditions are assumed within each stage (mixer
and settler). Therefore, the extract and raffinate streams leaving each stage
are at equilibrium. They take the indices of the stage they leave. The feed
is conventionally fed at stage N . The extract leaving stage N is the final
extract product.
In reality, the counter-current is realised using differences in density between
the phases.

200

6.4.2

Counter-current: stream definition

Before proceeding with the process design and the number of stages calculation for a counter-current column, it is needed to introduce a new stream,
called Delta. This will be used in the graphical construction as a kind of
operating line, in analogy to the McCabe-Thiele method.
Starting from the overall and component mass balances, the Delta is defined
as the difference of two streams exchanging solute at each stage.

Mass balances:
1. Overall
F + E0 = EN + R1 = M

(419)

Rj+1 + E0 = Ej + R1

(420)

2. From stage 1 to j

Applying the lever-arm rule to the overall mass balance, M can be positioned
on the diagram. In this case, the line connecting EN and R1 is not a tie-line,
since the two streams are NOT correlated by equilibrium.

201

Delta stream definition: From the mass balance equations, we get:


EN F = E0 R1 = Ej Rj+1 =

(421)

Delta is called difference stream or difference point. The Delta stream has
only the mathematical function of relating two streams exchanging solute at
each stage (Ej with Rj+1 ). In analogy to the McCabe-Thiele method, delta
can be seen as the operating line, hence used for the graphical calculation of
the number of stages. The position of delta is determined with the lever-arm
rule.

6.4.3

Counter-current: process design I

The purpose is the design of an extraction column. In the case of countercurrent layout there are two types of problem design (called I and II) depending on the given data.
In this first case, the solvent flow rate S is given as a data. In the following
example, the solvent is considered to be pure for the sake of simplicity.
202

In reality, the solvent will most likely come from a regeneration step and
contain traces of solute.
For the following procedure, we have:
F , X F , S, xS
DATA: F, XA
D
S
R, X R, X R
SPECIFICATIONS: XA
D
S
E , y E , y E , N, R
UNKNOWNS: EN , yA
1
D S

Step 1. Overall mass balance and lever-arm rule, determination of R1 ,


EN and yN .
The mixing point M1 is positioned from the mass balance
F + E0 = EN + R1 = M

(422)

In this case, the line connecting EN and R1 is NOT a tie-line, since the two
streams are not in equilibrium.

Step 2. Determination of the point.


Applying the lever-arm rule, the stream is identified on the diagram.
EN F = E0 R1 = Ej Rj+1 =

203

(423)

Step 3. Stage by stage: tie line through Rj , then find Ej , then find the
-line through Ej , finally find Rj+1 and repeat.
Based on the same principles as the McCabe-Thiele method, the number of
stages can be graphically determined using alternatively the tie-line (equilibrium) and the delta (operating line). In this example, 2 stages are needed
to reach the final extract concentration (E2 > EN ).

6.4.4

Counter-current: solvent rate and position: Smin and


Dmin

Most of the time in reality, the amount of solvent is an unknown (process


design II). Typically, the solvent flow rate is set by multiplying the minimum
amount of solvent to carry out the given extraction task by a factor.
Recall, in analogy with the McCabe-Thiele method (absorption and distillation), the minimum amount of solvent is the one which allows for the
separation with an infinite number of stages.
In the following example it is shown how to calculate the minimum solvent
flow rate by using a graphical method. The minimum solvent is linked by
204

the application of lever-arm rule to the position of a Dmin , which must be


then identified as a first step. The following example is only qualitative.
We start with the lever-arm rule to position M and as usual. If the
amount of solvent increases, M moves closer to the solvent vertex (M 0 ) and
delta moves closer to the diagram (0 ).

If the amount of solvent decreases, however, M moves in the direction of


the feed and delta moves further and further away from the diagram.

Decreasing the flow rate more and more, delta reaches an infinite distance
from the diagram when the lines SR1 and F EN are parallel. The flow rate
can be lowered further, until the slope of the F EN line changes and the
delta point moves to the right side of the diagram.

205

Finally, there is a minimum flow rate of solvent, Smin , with a corresponding


minimum . In analogy with absorption and distillation, at Smin the extraction is only possible with an infinite number of stages. Graphically, this
happens when the working line through EN F coincides with the equilibrium (tie-)line through F .
6.4.5

Counter-current: process design II

The purpose is again the design of an extraction column. In this second


case (process design II), the solvent flow rate S is unknown. Please check
below for the other data, specifications and unknowns for such a problem.
The example is only qualitative, and for the sake of simplicity, the solvent is
considered pure (although in reality it might still contain traces of solute).
For the following procedure, we have:
F , X F , xS
DATA: F, XA
D
S
R, X R, X R
SPECIFICATIONS: XA
S
D
E , y E , y E , N, R , S
UNKNOWNS: EN , yA
1
D S

Step 1.

Identification of min and solvent minimum Smin calculation.

206

Step 2.

Find final solvent flow rate


S = 1.2 1.5 Smin

(424)

Step 3.
I).

Overall mass balance and R1 , EN , yN determination (as for design

Step 4.

point determination (as for design I).

207

Step 5.

Stage by stage construction (as for design I).

208

Vous aimerez peut-être aussi