Previous Page

Chapter 4. DlSTILL4TION, ABSORPTION AND STRIPPING 153

Method Ideal area of application

Bubble point Narrow boiling range

Wide boiling range (non-condensables
Sum of rates present)
Excellent for absorbers and strippers
Average to wide boiling range
Regular mixtures
Tomich (or Holland variation)
Few trays
Suited to demethanizers and deethanizers
Large number of trays
Few components
Naphtali and Sandholm, Holland All types of thermodynamics
Chemical reactions
Initialization crucial
Few trays
Large number of components
Goldstein and Stanfield
All types of thermodynamics
Refining columns
Wide boiling range
Non-ideal and reactive mixtures
Boston and Sullivan
Suited to all column and specification
structures
Superfractionation, petrochemicals
Same as Boston
Russell
Refining columns, complex columns
Systems limited by mass transfer
Transfer model Non-ideal or reactive systems
Absorbers
Table

4.5.2 Dynamic Simulation of Distillation Columns

Dynamic simulation has appeared more recently and is becoming increasingly
popular as a tool in dealing with automation problems and process safety
studies, as well as in improving process design. In the specific case of distilla-
tion, this method has become indispensable for simulating, designing and
optimizing batch distillation units. The descriptive equations of tray operation
as defined by Figure 4.28 are expressed by Eqs. 4.48 to 4.51. They take liquid
154 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING

and vapor phase retention phenomena into account. Vapor phase retention is
usually disregarded.
Mass balance of component i on tray p :

- Vp+lYip+l - qJ-lXi,P.l - FP2i.P = 0 (4.48)

0 Energy balance of tray p :

d(Rbub) d(R:u:)
+ +(LP+SpL)hb +(VP+S,V)h,V
dt dt
- V p + l h p V + l - - L p l h k l - F p h ~ + Q p = O (4.49)
uL and uv are the internal energy of the liquid and the vapor (J/mol).
Equilibrium relation:

Yip+l (l - Ei.Mv.p) -k Ei.MV,pKi,pxi,p -Yi,p = (4.50)

Constraint equations:
Tb - TR,v= 0
Pb - P; = 0
1=n

i=1
c =1 (4.51)

This algebraicdifferential system of equations represents the retention

phenomena on the trays, considering extremely rapid mass transfer dynamics
in comparison with the other phenomena. In other words, the algebraic form
of the equation shows that the least fluctuation in composition at the tray inlet
has immediate repercussions on the composition of the exiting products. In
order to complete the description of the process, a number of equations must
be added to describe the operation of the condenser, the condensate drum,
the reboiler or the reboiler-retention vessel system depending on the chosen
technology:
The conventional approach in representing the dynamic operation of
heat exchangers consists in breaking down the apparatus into a number
of elementary cells so as to represent the hydrodynamic structure of
flows in the shell and the tubes. The descriptive equations of course
include the thermal inertia of the walls.
The simplest representation of the column’s retention volumes is that of
perfectly mixed tanks. This assumption is the most widely used in most
 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING 155

publications, but is not correct from a process standpoint. In fact, in most

condenser drums, the liquid stream has a longitudinal flow between the
point of entry and the weir allowing the reflux and the liquid distillate to
be drawn off. At the same time, the average pressure of the gas overhead
dictates the operating pressure of the column. In order to achieve a good
prediction, it is therefore necessary to be able to estimate the tempera-
ture and the composition of the interface. Recently, Latifi et al. (1996)
have shown the importance of this phenomenon. Figure 4.30 shows two
types of models representing a condensate drum, while Figures 4.31 and
4.32 describe the modifications in the dynamic response.
Since these vessels constitute the largest retentions in the distillation pro-
cess, it is clear given the results that the structure of the chosen model has a
great influence on the dynamic behavior of columns. Inversely, it complexifies
the model and increases the dimension of the problem to be solved. The user
must therefore look for the best compromise between accuracy and model
size according to his own objectives.
What holds true for vessels is also true for trays. Usual modeling consists
in considering the liquid and vapor phases as homogeneous, but as the sec-
tion on tray efficiency suggested, this ideal representation is insufficient to
characterize real phenomena. It would be necessary to approach a represen-
tation such as the one presented in Figure 4.33. However, in the specific case
of trays, it does not seem that complexifying the representation has a quanti-
tative influence on the overall model since liquid retentions are small and
there is significant interaction among all the stages.

Inlet

I Vapor I Vapor

L
Inlet

Figure
4.30 Models representing a condensate drum. d, stands for “perfectlymixed”.
156 Chapter 4. IS TILL AT ION, ABSORPTION AND STRIPPING

Time

Figure
Comparison o f variations in the condensate flow rate versus time, depending
on the type o f model chosen.

Hydrodynamic model

4
Hydrodynamic model
?!
3
c
-

f
-E
P

- Hydrodynamic model
E liquid outlet cell
2 -
-0
B
2c -
a,
z
6 -
Conventional model

Time

-
Figure
4.32 Comparison o f variations in condensate temperature at differentpoints of the
drum versus time according to the type o f model chosen.
 Chapter 4. DISTILLATION, ABSORPTION
AND STRIPPING 157

Conventional analogy Physical representation

IL

I"
Figure
4.33 Proposed models representing trays.

Other equations can also be added to describe the process completely. The
most important ones involve the controllers installed on the unit (Eq. 4.52).
Others too can give the influence of liquid and vapor flow rates on pressure
drop and retentions. Equations are usually algebraic (see Eqs. 4.53 and 4.54)
since the dynamics can be considered to be very fast:

U applied order variable

Uref reference order variable
y real value of the controlled variable
yspecspecified value for the controlled variable
Kc proportional constant
zl integral time constant
T~ derived time constant

Equation for tray pressure drop

Pk - Pk+1+Giq [ ( L N + Sh) 0LNI 2/3 + G a p (VN+1 uk+J = 0 (4.53)
where C,, and Cvaprepresent the pressure drop coefficients due to the liquid
and the vapor. They come from the tray pressure drop equations. u h and u ; + ~
indicate the molar volumes of the liquid and the vapor phase respectively.
Equation simulating retentions on the trays:
R f i ~ -h Rh,dnuk - C1iq[ ( L N + Sh) uL,] 2/3 - Cvap(VN+1u",J =0 (4.54)
where Rk,dn is the captive retention on the tray.
158 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING

A number of mathematical methods can be used to integrate the system of

algebraicdifferential equations (Runge-Kutta method, Powell method, Gear
method, etc.) in order to solve this very large dimension system for a distilla-
tion column. The most widely used is the Gear method, which is ideally suited
to the characteristics of the equations to be solved. In other words it is suited
to the simultaneous presence of phenomena with rapid, or even instantaneous
dynamics and those with slow dynamics.
Commercial software packages are available to carry out such simulations.
Designed in a way similar to static simulators, they can access thermodynamic
libraries, unit operations, mathematical tools, etc. and can be used to take
equations written by the user into account. Generally speaking, they are inte-
grated in or interface with the structure of static simulators and make use of
their technical possibilities. There seems to be a much smaller supply of this
type of software today than for static simulation. Nowadays, the three main
suppliers offer dynamic versions of their software:
SPEED UP by Aspen Tech (USA)
PROTISS by Simulation Sciences (USA)
HYSYS by Hyprotech (Canada)
A fundamental comment remains to be made about dynamic simulators: All
the preceding equations describe phenomena with local parameters, i.e. all
the descriptive variables have a single value in the whole volume under con-
sideration. This is one of the reasons why analogies have been proposed to
describe hydrodynamic phenomena better. The same problem arises with
packing. The phenomena involved are continuous all throughout the column.
A strict representation requires the use of equations with partial derivatives
(in relation to time and to the space variables, i.e. the longitudinal and radial
position in the bed). In the present situation, dynamic process simulators can
not do this readily. To achieve a strict representation, solvers are required that
are better suited, such as those employed by hydrodynamic simulation tools.
Figures 4.34 and 4.35 give results of dynamic simulations for crude oil dis-
tillation.

4.6 Designing a Distillation Separation Unit

Earlier we analyzed the design of a column and proved that for a column taken
individually there was an optimum number of trays for specified economic
conditions. In fact, the column is always integrated in an energy context, or
rather in a cascade of distillation columns, since it is common for the separa-
tion problem to be more far-reaching, to consist in separating a feed into sev-
eral products. In this case it is indispensable to define the order in which
separations will be carried out before sizing each of the columns.
LIVE GRAPH
Click here to view
Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING 159

0.62

0.60

0.58 A

0.56

0.54

0.52

0.50

0.48 I I I I I I I I *
5 10 15 20 25 30 35 40
Time (lo3s)

165 -

Dynamic simulation for crude oil distillation.

A. L/V ratio top tray
B. 95%point ASTM 086 of gasoline.
 LIVE GRAPH
Click here to view
160 Chapter 4. DISTILLATION,
ABSORPTION AND STRIPPING

130

.
c
128

126
A
$2
!$

--s
3 124
0

122
9
u)

?? 120
.-0
z
‘p 118
In
116
2
114

112

110
5 10 15 20 25 30 35 40
Time ( 1 0 ~ s )

311

310

309 B
308

307

306

305

304

303

5 10 15 20 25 30 35 40
Time (1O3 s)

Figure
4.35 Dynamic simulation for crude oil distillation..
A. Atmospheric residue flow rate.
 Chapter 4. DlST/LLATlON, ABSORPTION AND STRIPPING 161

4.6.1 Reminder of the Pinch Method

Let us consider a process where there are a number of hot fluids, i.e. those to
be cooled down and a number of cold fluids, i.e. those that have to be heated
up so that they can be used. To illustrate this concept, the main hot and cold
fluids in a binary distillation column are given below:

Hot fluids Cold fluids

Overhead vapor Reboiling

Distillate
Residue

After establishing this balance and adding the hot and cold process utili-
ties, it becomes easy to construct the cumulative temperature curve accord-
ing to the power available in the hot fluids and the power required to heat up
cold fluids. Constructing composite curves leads to Figure 4.36. It means con-
sidering the existence of a countercurrent heat exchange between a pseudo
hot fluid, the sum of all the hot fluids, and a pseudo cold fluid, the sum of all
the cold fluids. This type of hypothesis defines the most efficient transfer
mode, and therefore the mode that requires the minimum heat exchange area.
From a practical standpoint, ideal actualization of this concept yields exchang-
ers whose representation is vertical in the diagram. Inversely, any existing

Surplus
heat to be
eliminated
Heat exchange
between process fluids
/ Heat to be
supplied

/
Figure
4.36 Example of a temperature-enthalpy diagram.
162 Chapter 4. DISTILLATION, ABSORPTION
AND STRIPPING

exchanger whose representation is greatly different from the vertical exhibits

faulty design and therefore an overinvestment.
The diagram can be broken down into three zones: the one in the middle
corresponds to heat exchanges between process fluids, the one on the left r e p
resents the amount of heat that has to be eliminated by cooling equipment
while the one on the right defines the amount of heat to be supplied by energy
vectors outside the process.
It can be seen that there is a point where the difference in temperature
between hot and cold fluids is minimal. The existence of such a point, called
the pinch point, has two major consequences:
1. The temperature difference at the pinch point defines the economics of
the energy management system. Even if we attempt to add an exchanger
in the zone with wide differences in temperature, the real profitability
depends on the pinch point gain.
2. The pinch point exists because to its left the process has surplus calo-
ries. To the right there is a deficit. This concept is fundamental and gov-
erns any and all attempts to improve a process. The objective of any
modification is to provide new needs in the surplus zone where energy
is free of charge and new availabilities in the right-hand zone of the dia-
gram in order to minimize the deficit.
A typical example of application is in defining the conditions to be com-
plied with in order to integrate a compression heat pump in a process. The
heat pump is made up of a compressor which imports high-level energy, an
evaporator which absorbs the heat (cold fluid) at the temperature Tevap and a
condenser which exports heat (hot fluid) at the temperature Tcond.
Performance is characterized by the COP, coefficient of performance, defined
as the ratio between the power of the condenser and that of the compressor.
It is useful to compare the COP to the coefficient of performance for the pro-
cess unit (COU), i.e. the gain that is really obtained in the unit divided by the
power of the compressor. This balance is summarized in Table 4.5.
As a reminder, any cold fluid added in the calorie deficit zone increases the
energy requirements and any hot fluid added in the calorie surplus zone
increases the power discharged through the cooling equipment:

Tevap < Tcond < Tpinch cou =

< Tpinch < Tcond cou = cop
Tpinch < Tevap Tcond cou =
- Table
4.5 Comparison of COP and COU.

It can be seen in Table 4.5 that the only conditions in which the pump’s per-
formance is worthwhile are when the condenser and evaporator are located
on either side of the pinch point. This is absolutely logical since this design
means adding a hot fluid in the deficit zone and a cold one in the surplus zone.
 Chapter 4. DISTILLATION, ABSORPTION
AND STRIPPING 163

The preceding condition does not assume profitability of the total investment,
which includes the pump and the extra exchange area needed for process heat
transfers.

4.6.2 Integration of a Distillation Column in a Process.

Choosing Pressure
Integrating a distillation column in a process is considered in the same way. In
this case, the column is limited to a hot fluid, overhead vapors, supplying
Qc (kW) at a temperature Tc , and a cold fluid, the reboiler, consuming QR (kW)
at the temperature TR. The different scenarios that can be considered for inte-
grating the column are given in Table 4.6 which expresses the overall balance
for the unit.

Variation in overall
Variation in
process energy
surplus calories
demand

-
Table
4.6 Integrating a column in a process.

Accordingly, the worst scenario is obtained when the condenser and

reboiler are on either side of the pinch point. In this case, the column con-
sumes heat in the deficit zone and exports it to the surplus zone. The only way
to offset this situation consists in installing a heat pump or vapor compression
to recycle the energy. It can also be seen that the conditions stated previously
are thoroughly complied with.
The two other configurations are of interest since the column’s influence
on the operation of the zone under consideration depends on the difference in
thermal load between the reboiler and the condenser. This difference varies
depending on the enthalpy balance of the column and on the feed preheating
conditions, among other factors.

4.6.2.1 Choosing Pressure

The only means available to make the reboiler and condenser thermal levels
vary is the pressure. However, an increase in operating pressure has the fol-
lowing consequences on the column as a general rule:
1. The relative volatility decreases, thereby making separation correla-
tively more difficult. For a set number of trays, the energy consumption
goes up.
164 Chapter 4. ~ISTILLATION,ABSORPTION AND STRIPPING

2. The reboiler thermal level may be such that fouling and polymerization
occur and cause product degradation and added operating costs.
3. The vapor volume flow rate decreases as the pressure increases. This
variation reduces column diameter. Above 10 bar, the pressure requires
a shell ring thickness which adds extra expense to the process. Below
atmospheric pressure, the fact of placing the unit in a vacuum causes
problems with strength of materials and fluid tightness, as well as heavy
costs.
As a result, when there are degrees of freedom the pressure is adjusted so
that the condenser and reboiler are on the same side of the pinch point. When
constraints are such that this variation is impossible, pressure is determined
by the cooling fluid (air or water) and the allowable difference in temperature
for these exchangers.

4.6.2.2 Intermediate Reboilers and Condensers

As suggested in the section on calculation methods for multicomponent
columns, it is possible to import or export heat on each of the trays in the col-
umn. As a reminder, the rectifying zone has surplus calories so any heat
import becomes counterproductive in terms of separation. The analysis is of
course the reverse in the stripping zone. It may be useful to add intermediate
reboilers or condensers though. Some authors such as Rivero et al. (1991)
have even thought of adding this type of exchanger at each stage to minimize
exergy loss in the column. Reduced exergy loss is in fact due to the modifica-
tion of heat transfer thermal levels, whereas total consumption is on the
increase. This phenomenon is easy to understand by referring to the example
in Figure 4.37. The separation duty of a tray in the stripping zone is an increas-
ing function of the V/L ratio. By adding an intermediate reboiler, the ratio is
modified in the lower zone and becomes V- AL/L - AL, less than V/L. In order
to offset the separation duty loss on the lower trays, the reboiling power and
reflux ratio must be increased again.
Figure 4.38 illustrates this giving an example of variation in energy con-
sumption versus intermediate reboiler power. These conclusions must not
cause the usefulness of this technology to be called into question. If the inter-
mediate reboiler thermal level is such that it is situated in the surplus calorie
zone, the operating cost in fact decreases.
The same type of analysis can be made for intermediate condensers. They
are advantageous when the thermal level is such that calories are exported to
the deficit zone. Industrial applications of such concepts are frequent, for
example the circulating reflux techniques used on crude oil atmospheric and
vacuum columns. They are used to recover calories that can be useful to pre-
heat the crude.
Intermediate condensers are also widely used in cryogenic columns. They
reduce the average thermal level at the top of the column, thereby reducing
operating costs.
 Chapter 4 DISTILLATION, ABSORPTION
AND STRIPPING 165

I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I

z xD

Figure
4.37 Intermediate reboiler and condenser.

A
9 -
8 -
7 -
6 -
5 -
4 -
3 -
2 -

w
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Fraction of total power provided by intermediate reboiler

Influence of intermediate reboiler power on total energy consumption in

gasoline stabilization.
166 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING

4.6.2.3 Vapor Compression or Heat Pump

The economic advantage of these devices depends of course on the relative
value of energy vectors. It is a good idea to discuss them whatever the present
context may be, because they are part of the range of usable technologies.
Three main set ups can be envisaged (Figs. 4.39a, b, c):
Diagram I uses a heat pump. The calories available at the condenser
vaporize the working fluid which is then compressed to increase the pro-
cess fluid’s condensation temperature. Thermal energy is reused in the
reboiler. Compared to the vapor compression solution, the drawback of
this one is lower performance because of the two exchangers. The advan-
tage is that it does not work directly with the process fluid, thereby avoid-
ing any corrosion problems or the use of a non-standard compressor.

A
f
I+ Distillate

Compression
\ 7 ‘Expansion
J ,

I
&
‘r‘ Residue

1 Figure
4.39a Diagram I: heat pump.

Diagram I1 uses vapor compression. The column’s overhead vapors are

compressed so as to meet the reboiler’s thermal needs. The advantages
and drawbacks have already been discussed.
 Chapter 4 DISTILLATION, ABSORPTION AND STRIPPING 167

However the two techniques may be implemented, they have the draw-
back of connecting the condenser and reboiler thermally. Performance
(COP), expressed by the ratio between useful energy and compressor
power, depends heavily on the temperature difference between top and
bottom. In an initial approximation, the COP is defined according to the
Carnot theoretical efficiency by Eq. 4.55:
m

COP = 0.5 - 1U
(4.55)
TR - Tc
The gain potential is thus rapidly quantified, the minimal performance
threshold corresponds to the case where the performance coefficient is
higher than the ratio between the price of electric and thermal power. The
maximum temperature difference is around 50°C in the present context.

Compression

Residue

L
Figure
4.3913 Diagram 11: vapor compression.

Diagram 111 corresponds to a totally different approach. The liquid com-

ing from the bottom tray in the column is expanded through a valve
where it is partially vaporized. Since the vaporized fraction depends on
168 Chapter 4. DISTILLATION, ABSORPTION AND STRlPPlNG

the expansion rate, added heat is often needed to ensure column opera-
tion. The compressed vapors are reinjected in the column. The advan-
tage of this technology, compared to the previous solutions, is that the
column’s reboiler and condenser are decoupled and conditions of appli-
cation become wider. We come up against the general problem discussed
with the pinch method once again: this type of modification will be e c e
nomically advantageous only if it allows the free calories available in the
system to be utilized.

-
Figure
4 . 3 9 ~ Diagram 111: decoupling the reboiler and the condenser.

4.6.3 Estimating Investments. Seeking Optimum Sizing

All of the technical solutions presented earlier were put forward solely in terms
of minimizing operating costs. The analysis must be supplemented by an esti-
mate of investments in order to identify the best overall economic solution.
Chauvel et al. (1976) have proposed a simplified investment estimation
methodology whose accuracy is approximately 30%. This degree of accuracy is
sufficient for a draft project when process options are compared. The principle
 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING 169

of the method is to determine a reference cost by means of a correlation based

on an equipment characteristic (weight for the column, diameter for trays and
area for exchangers) and then to apply corrective factors to take material con-
straints and installation costs into account.
We suggest the reader refer to Volume 4 of this collection for more detailed
indications on sizing tray and packing type columns, reflux drums, reboilers,
condensers and heat exchangers as well as estimating the costs of these
equipments.

4.6.4 Controlling a Distillation Column

A distillation column in operation undergoes a number of perturbations due to
the feed composition, the feed flow rate, the outside temperature, the modifi-
cation of the reboiler fluid’s thermal level, etc. Even if the retention vessels
placed at the top and the bottom of the column attenuate fluctuations, regula-
tion structures are necessary to correct the column’s operating conditions.
The most commonly admitted regulation structure uses a series of mono-
variable controllers (F’ID) whose set points can be modified either by a series
of controllers placed upstream or by an optimizer that takes the real process
variations and the set point changes into account. It is therefore imperative to
know how to install control loops to manage the degrees of freedom of the sys-
tem, adjust the controller’s parameters and find efficient solutions for the opti-
mizer in order to deal with this problem.
For a continuous distillation column the major variables that need to be
controlled are:
the pressure;
the distillate flow rate;
the reflux flow rate;
the residue flow rate;
the reboiler fluid flow rate;
the level of top and bottom drums.

4.6.4.1 Controlling Pressure

a. Liquid Distillate
Pressure is usually controlled on the basis of the condensate drum tempera-
ture, which sets the vapor pressure and, by means of all the pressure drops,
sets the pressure of the column. In order to achieve this variability in temper-
ature, the condenser is oversized and the thermal load varies during opera-
tion. Two technical solutions are available (Fig. 4.40):
1. The first consists in having the cooling fluid flow rate vary. For a water-
cooled condenser, the flow rate is controlled and the PC bressure con-
troller) acts on orders from the FC (flow controller). For an aircooled
system, the rotation speed of the blades regulates the air flow rate
170 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING

(Fig. 4.40A). Let us note that the solution of installing a battery of air-
cooled heat exchangers with only one having a variable speed is often
preferred. During operation, the number of aircooled exchangers work-
ing is chosen so as to ensure flexibility with the one having variable
speed.
2. The second consists in installing a cooling system by-pass (Fig. 4.40B).
Modifying the vapor flow rate going through the by-pass modifies the
temperature.
In some conditions where the distillate is highly volatile, a split-range sys-
tem is used (Fig. 4.409. Beyond a maximum threshold, the condensate drum
is degassed and the pressure is controlled by the gaseous distillate technique.
In the normal zone, the pressure fluctuates freely according to the quality of
overhead vapors. Below the minimum threshold, the unit is regassed by
importing high pressure gas. This technology is very often used on crude oil
atmospheric distillation units. The regassing fluid is the fuel gas available on
site in this case.
b. Gaseous Distillate
In this instance the gaseous distillate is used to control pressure by adding a
flow control valve (Fig. 4.41).
c. Atmospheric Column
For columns working at atmospheric pressure, control is achieved simply by
connecting the column with the atmosphere through a duct leading to a vent.
This choice of course assumes that the distillate quality is such that losses of
matter are extremely small.
d. Vacuum Column
Steam ejectors and/or vacuum pumps can be used to create a vacuum in a dis-
tillation column. From an economic standpoint, it is difficult to select the best
technology blindfolded. Although vacuum pumps certainly require larger
investments, operating costs depend on the energy equilibrium of the facility
and the relative price of HP steam and electricity. The degree of vacuum
achieved by a vacuum pump depends on the characteristics of the machine
and on the flow rate of gas to be withdrawn. For ejectors, it depends on the
number of stages installed and the downstream condensation temperature
(and therefore on the temperature of the cooling water).
e. Floating Pressure
Reducing pressure in a column usually has a positive effect on separation per-
formance, thereby enhancing product quality or reducing reboiler power and
reflux ratio to maintain specifications. When the cooling fluid undergoes wide
temperature fluctuations with time, it becomes advantageous to modify the
pressure setting in order to exploit the cooling resources to a maximum and
achieve the minimum possible pressure at all times. This approach, called
floating pressure control, significantly reduces energy consumption over a
long time frame. Some authors state a 10%gain on operating costs.
 Chapter 4. DISTILLATION,
ABSORPTION
AND STRIPPING 171

,Ir
A

I 2
B

Gas Flare

7 1 C

Figure
4.40 Differentcontrol diagrams for the pressure at the top of the column in the case
of a liquid or mixed distillate.
172 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING

Figure
Differentcontrol diagrams for the pressure at the top o f the column in the case
of a vapor distillate.
 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING 173

4.6.4.2 Basic Control of a Distillation Column

At the top and bottom of the column one of the two available fluids (reflux/dis-
tillate or residue/reboiling fluid) must be used to control the liquid level in the
corresponding vessel. Using the reflux for overhead control recycles into the
column the small perturbations occurring in the condensate drum. This is not
a sound situation since internal column flows are not properly governed and
cause a wide fluctuation in product quality. This is why the configuration pre-
sented in Figure 4.42 is preferred. The distillate and the residue control levels
while the reflux and reboiler fluid flow rates are controlled in such a way as to
stabilize conditions inside the column.

Condenser

-
Figure
4.42 Basic control of a distillation column.
174 Chapter 4. DlSTlLlATlON, ABSORPTION
AND STRlPPlNG

Constant vapor and liquid flow rates are favorable to column operation, but
do not automatically guarantee good performance for all sorts of perturba-
tions. Table 4.7 presents a number of scenarios.
In many circumstances such consequences are not admissible because
they affect product quality too severely. To correct these defects, control set-
tings must be modified according to the events that occur in the process.

Consequences Consequences
Perturbation
on the distillate on the residue
Increase in feed flow rate Slight modification in dis- Increase in residue flow
tillate flow rate depend- rate along with the quan-
ing on enthalpy balance tity of light products
Increase in light product
concentration
Increase in inlet tempera- Increase in distillate flow Decrease in residue flow
ture rate and therefore in rate and therefore of the
heavy product concen- residue’s light fraction
tration
Increase in light product Constant distillate flow Constant residue flow
concentration in the rate rate
feed Increase in light product Increase in light product
concentration in the dis- quantity in the residue
tillate
Increase in heavy product Constant distillate flow Constant residue flow
concentration in the rate rate
feed Increase in heavy product Increase in heavy product
concentration in the dis- quantity in the residue
tillate

I I 4.7
Consequences of a number of perturbations involving column inlet.

a . Insensitivity to Feed Flow Rate

Assuming constant feed composition, a distillation unit’s separation perfor-
mance is readily kept constant by making basic control settings (reflux flow
rate and reboiler fluid flow rate) proportional to the feed flow rate. In order to
achieve this, a flow sensor placed on the feed adjusts the set points of local
controllers.

b. Controlling Product Quality. Using a Sensitive Tray

If we assume the hypothesis of the theoretical tray, temperatures are equal to
liquid bubble point temperatures, which are dependent on composition and
pressure. Inversely, the temperature represents the composition, except for
 Chapter 4. DISTILLATION, ABSORPTION
AND STRIPPING 175

binary mixtures where it measures the composition. By controlling the tem-

perature of one of the trays, the quality of one of the products is influenced
without, however, strictly guaranteeing it is constant. The fundamental prob-
lem during implementation consists in identifying the tray whose temperature
variation is the most sensitive to operating modifications. It is identified exper-
imentally or by simulation as shown in Figure 4.43.

Rectifying sensitive tray

/ Stripping sensitive tray

il
c

c
U L a y s
E
b
Z

Figure
Example of identifying a sensitive tray. Variation of temperatures to variations
+ 4 - A of product composition.

The simplified flowsheet showing sensitive tray control is given in

Figure 4.44. Part A applies to the rectifying zone and part B to the stripping
zone. This type of flowsheet calls for two comments:
1. It ensures constant quality of one of the two products (distillate or
residue), with the other varying according to column perturbations.
2. When the column’s pressure varies significantly with time, the value of
the setting must be revised to keep from affecting the quality of the rel-
evant product. This is done by applying a pseudo-Arrhenius equation
around the reference point.
This type of structure is absolutely satisfactory when it is possible to “sac-
rifice” one of the two products. In order to maintain the quality of both at the
same time, two controls would be required per sensitive tray, one applying to
the rectifying zone and the other to the stripping zone. This solution can not
be contemplated, as applying contradictory top and bottom control settings
would cause instability. A “referee” is needed to handle setting changes and
restore column operation as soon as possible.
176 Chapter 4. DISTILLATION, ABSORPTION
AND STRIPPING

-
Figure
4.44 Sensitive tray control.
A. Rectifyingzone. B. Stripping zone.

The referee is usually a model which, when implemented, makes it possible

to use a conventional methodology in automation: control by internal model
whose operating principle is given in Figure 4.45.
The status of the column is estimated with temperature measurements in
the stripping and rectifying zones, and flow rate measurements for distillate
and residue. It is regularly recalibrated with measurements of the column's
product composition. Based on this, a management policy for set points
is defined for a specified time frame in order to restore normal operating
 Chapter 4. DISTILLATION, ABSORPTION
AND STRIPPING 177

Reboiler power
1-
Reflux flow rate
Distillation
Sensitive temDerature in rectifying
Sensitive temperature in stripping
Distillate flow rate
Residue flow rate ,
Distillate composition
Residue composition
I I I ' I I I1

I Internal model
I
1 Figure
4.45 Control by internal model.

conditions. It is applied between two data acquisitions and revised at that

time. Control quality depends greatly on the quality of the model used.
The technique of Dynamic Matrix Control uses a dynamic linear model tak-
ing interactions among variables into account and identified experimentally
around an operating point. This type of approach is limited by the linear
model's performance. As soon as operating conditions get too far from the
conditions of the nominal point, the model's predictions are mistaken and con-
trol quality degenerates. In order to overcome this limitation, Levine and
Rouchon (1991) have suggested using a simplified non-linear model to take
account only of dynamics of interest to control. This approach has resulted in
the COLBIN software (Laforet and Djenab, 1992) whose numerous applications
in the field have given convincing results. Figure 4.46 gives an example of an
experimental response obtained in this way for a debutanizer.
c. Regulating Controller Parameters
The parameters of PID controllers (proportional, integral and derivative con-
stants) depend on the characteristics of the dynamic response to a perturba-
tion. Corriou (1996) has put different controller regulation protocols together
and proposes those in Table 4.8 among others. They utilize a number of param-
eters identified on the step response curve in relation to the action variable:
K is the static gain identified as the ratio between B and the value of exci-
tation amplitude A
z is a measurement of the response dynamics defined as the ratio between
the value of the slope at the downdip and the amplitude of the
response B;
tD is the delay of the response.
In the special case of a distillation column, the dynamic characteristics
vary and depend on the specifications laid down on separation. Table 4.9
quantifies these variations. In order for the quality of column control to be
constant, controller adjustments must be modified according to running
conditions.
178 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING

% C4 bottom

/
Set points Measurement
Yo cs top

6 12 18 24 30

200

Bottom temperature (“C)

150 -

Feed flow rate (t/h)

100
6 12 18 24 30 36
 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING 179

Type of Proportional Integral time Derivative time

controller constant constant constant

K,= -
1 - 7( l + $ )
P
tD

Dt
30+3-
PI
1 7
K,= - - (0.9
tD
+$ ) ?

D
t
32+6-
PID

1Tabler
I I
4-8
PID regulation parameters recommended by Corriou (1996).

AXD
Static gain - .lo6 Time constant (min) Delay (s)
AV
584.2 24 =O
358.8 16 =5
272.5 13 = 10
225.2 9 = 15
194.6 8 = 20

Variations in the transfer function linking residue composition to vapor flow

rate at the bottom of the column. Application to an 18-tray benzene-toluene
separation column according to product quality (Mac Donald and Mac AVOJ
1987).

4.6.5 Designing a Separation Flowsheet

Very often a separation unit incorporated in the process separates the stream
into several products by means of several columns. Where the columns are
located in the flow diagram will have a very great influence on the economics
of the separation unit. Some authors have in fact shown that, if investment
depends mainly on the feed composition and the expected product specifica-
tions, operating costs depend on the structure of the flow diagram. Case stud-
ies have shown that consumption varies by a factor of two depending on the
flowsheet chosen.
180 Chapter 4. DISTILLATION,
ABSORPTION AND STRIPPING

It would be much too complex to find the best flow diagram by systematic
study, since [ 2 (p - l)] ! / p !(p - l)! structures exist for p products. It is indis-
pensable to use a few heuristic rules or some more complex process synthe-
sis methodologies as a basis for accelerating the search and minimizing costs.

4.6.5.1 Heuristic Rules

The rules defined below are based on distillation characteristics. They give
guidelines for organizing separation flow diagrams, but their general nature
may make them incompatible with a specific mixture. Outside of an applica-
tion context it is impossible to rank them. The engineer will have to look for
the best compromise.
1. First eliminate majority components.
2. Start out with easy separations.
3. Give priority to balanced separations, i.e. where distillate and residue
flow rates are of the same order of magnitude.
4. End up with separations that cause problems (azeotropic distillations
for example).
These rules give the guidelines for defining flow diagrams. However, the
fact that they can not be ranked often means decisions are made on a case by
case basis. For example, what is the best flow diagram for separating a mixture
of n-butane (lo%), isopentane (60%), n-pentane (25%) and n-hexane (5%)?
In order to clarify the problem, let us assume that all the columns perform
separations between adjacent substances in the scale of relative volatilities.
We know that distillation conditions (reflux ratio and reboiling ratio) depend
solely on the relative volatility between the key substances and on their par-
tial flow rate. In contrast, energy consumption is at first glance proportional to
the term (1 + r). It thus depends on the partial flow rate of the light distributed
component, but also on the partial flow rate of nondistributed components.
Consequently, in order to minimize energy consumption in the flow diagram,
the sum of the nondistributed component flow rates must be reduced. If we
refer to Figure 4.47, the sum is equal to F, (partial flow rate of component C in
the feed) for diagram I and to FA (partial flow rate of component A in the feed)
for diagram 11. By applying this rule to the preceding problem, the best flow
diagram would consist in performing the Cpentaneln-pentane separation first,
then separating the two binary mixtures (n-butane/i-pentane and n-pentane/
n-hexane).

4.6.5.2 Simplified Methods

Let us consider the special case where the mixture has to be separated into
three cuts, independently of the number of components actually present in the
mixture. A number of flow diagrams can be contemplated to deal with this
problem. Tedder and Rudd (1978) have proposed seven flow diagrams shown
together in Figure 4.47.
 Chapter 4 DISTILLATION, ABSORPTION AND STRIPPING 181

I II
IA

IA

Ill IV
IA

-A IA
V
?- VI

+B

i-I
VII

Figure
Multicomponent mixtures. Seven possible separation diagrams (Tedder and
Rudd, 19 78).
182 Chapter 4. DISTILLATION,
ABSORPTION
AND STRIPPING

Diagram I is the direct cascade where the product is topped in two

columns. The presence of four degrees of freedom, the reflux ratio and reboiler
power for each column yield all specifications. Diagram 11 or the inverse dia-
gram has the same characteristics. In diagram Ill a vapor sidedraw is rectified
in a column with its own reflux flow rate. This structure provides the same
number of degrees of freedom (vapor flow rate and reflux flow rate), thereby
yielding all specifications. The same holds true for diagram IV where product
B is obtained by stripping a liquid offtake from the principal column. In dia-
gram V, the first column separates the feed into two cuts which serve as feed
for the column downstream, producing the light cut at the top, the median cut
on a sidedraw and the heavy cut at the bottom of the column. Provided that
separation is kept under control in the first column, any specification can be
achieved in the downstream column. As shown in Table 4.10, this remains true
for the median component. High purity is therefore achievable provided that
the distillate of the upstream column is stripped of heavy product and the
residue is inversely stripped of light product. This is logical because the flow
diagram has the same number of degrees of freedom as the preceding flow dia-
grams. Diagrams VI and VII are simplified alternative solutions since product
B drawn off from a sidedraw is not reprocessed in a downstream column. In
diagram VI, B is necessarily polluted by A and in VII by C .

Preseparation Final column

I

1 Feed Distillate Residue Distillate Offtake Residue

lsobutane (%rnol) 39.97 0.03 99.5 0.12 0.0
n-Pentane (%rnol) 59.93 60.07 0.5 99.66 0.5
0.10 39.90 0.0 0.22 99.5
Flow rate (kmol/h) 50 50 20 60 20

M Example of fractionation obtained with diagram V

Figure 4.48 gives the best flow diagram according to the feed composition
which is represented by a single point. When the separation index, defined by
the product am aBC (where am and aBC represent the relative volatility of the
key components characterizing separation AB and the inverse of the relative
volatility characterizing separation BC), is less than 1.6 (difficult separation),
the selection chart is defined by Figure 4.48A. It is defined by Figure 4.48B
when the index is higher than 1.6. These results are extremely important,
because they can be used to get the best structure very quickly. Let us note
that the boundaries limiting each of the zones have been defined from corre-
lations established on the basis of case studies. As a result, they are not strict
and, for any point characteristic of the feed in the neighborhood of these
curves, it is useful to check how well adjacent solutions fit the bill.
 Chapter 4. DISTILLATION. ABSORPTION AND STRIPPING 183

B A

Multicomponent miwtures. Selecting the best flow diagram depending on the

(4481 value of the “separation index” (Tedder and Rudd, 1978).
A. Index C 1.6. B. Index > 1.6.

Illustrating the use of these charts, Figure 4.49 shows application t o the
stabilization of wide range gasoline coming from crude oil distillation. The
most commonly accepted flow diagram consists in stabilizing the wide range
gasoline before separating it into light gasoline or heavy naphtha. As shown in
the figure, the inverse diagram can be more efficient in certain flow rate and
184 Chapter 4. DlSTlLLaTlON, ABSORPTION
AND STRIPPING

- Q P L
Light gasoline

n
. LPG

Light gasoline

--
Figure
4.49 Comparison OF two light gasoline-heavy naphtha separation flow diagrams.

composition conditions. Figure 4.50 gives another example where the separa-
tion diagram separates a light component present in a very large quantity and
a heavy substance present in a small quantity. The structure presented seems
satisfactory at first glance since it follows the heuristic rules stated earlier and
proposes obtaining the lighter substance in two steps. The first column repre-
sents an easy separation, the fact of cutting right in the middle of A readily
gives a highly pure distillate. The second column which meets the separation
specifications thus minimizes losses of product B. However, if we refer to the
Tedder and Rudd diagram for the downstream part of the separation, we can
 Chapter 4 DISTILLATION, ABSORPTION AND STRIPPING 185

A A
-Q

Example of separation flow diagram with the light component largely in excess
compared to the heavy component.

see that the best structure is diagram V. Using it and optimizing the operating
pressures so as to make thermal connections between reboiler and condenser,
we arrive at the final flow diagram in Figure 4.51. It allows consumption to be
reduced by approximately 35% compared to the initial diagram.

4.6.5.3 General Methods

The ideal solution would be to have software which, based only on knowing
feed composition, would choose unit operations, find the best flow diagram
structure and optimize sizing. Although a lot of research has been done in this
area (mainly on heat exchanger network design and distillation separation
flow diagrams), this ideal product does not exist. Two major approaches are
proposed: the first uses the principle of distillation paths or residue curves
which have proven their worth for azeotropic separations (see
Section 4.9.2.2), and the second employs mathematical optimization methods,
generally of the "Mixed Integer Non-linear Programming" type. Binary vari-
ables can be used to accept or reject the existence of a unit, and continuous
variables can be used to optimize the flow diagram. In order to do this, first of
all a very general process structure (conventionally termed superstructure)
must be defined where all the alternative solutions are included. The algo-
rithm ultimately proposes the best solution. Nishida et al. (1981) present an
excellent review of all these process synthesis methodologies. In their present
state, their usage is limited by a number of restrictive hypotheses: the
186 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING

-
F

I R

Figure
Optimized flow diagram of the case in Figure 4.50.
A. Variation in the feed point during optimization of the distillate flow rate.
B. Optimized flow diagram.
 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING 187

columns must be simple (one distillate and one residue) and the separations
must be clear cut (recovery ratios close to 1). Additionally, they can not read-
ily be utilized to take thermal integration between condensers and reboilers
into account. This latter aspect is not too detrimental, as demonstrated by
Smith (1995), the best flow diagrams without thermal integration often turn
out to be the best after thermal integration. In the opposite case, the differ-
ences are small. Flow diagram synthesis software exists and can be used effec-
tively for difficult problems. When no such products are available, the
following methodology can be used. Based on a dynamic optimization tech-
nique, i.e. moving back toward the feed after having determined final separa-
tions, it can be used to decide among the different heuristic rules. Let us
consider the example of Table 4.1 1 extracted from technical literature. It con-
sists in separating a feed of five components. All the final binary separations
are the AB, BC, CD, DE separations for which it is possible to find the optimum
cost based on short cut methods or tray models. The ternary separations that
can be contemplated are ABC, BCD, CDE and for each one the total cost is

Optim~mdown-
Subgroup Separation Total cost
stream separation

AB A/B 0.2613
BC B/C 0.9433
CD C/D 0.5927
DE D/E 1.692
ABC A/BC 1.3446
-/c 1.4593
BCD B/CD 1.7187
BCP 1.7168
CDE C/DE 2.4737
CD/E 2.4457
ABCD A/BCD 2.1875
AB/CD 2.2590
ABC/D 2.2891
BCDE B/CDE 3.7797
BCPE 3.5896
BCD/E 4.1358
ABCDE A/BCDE 4.1571
AB/CDE 4.3570
ABC/DE 4.1856
ABCD/E 4.8474

Differentseparations For a mixture OF Five components.

(411
188 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING

equal to the optimum cost of the ternary separation itself and of the down-
stream separation. For example, the best separation of the ABC ternary will be
identified by comparing the cost of the A/BC step followed by B/C with the
AB/C step followed by A/B.
The procedure is then applied in the same way to the quaternary mixtures
and so on until the feed is reached.
This limitative procedure, given the initial assumptions:
minimizes the number of calculations by structuring the design
approach;
helps decide, if necessary, between the different heuristic rules:
gives a satisfactory flow diagram structure, even if taking real specifica-
tions into account means that the operation of each of the columns is
modified later on.
However, it does not consider the possibilities of thermal integration
between the reboiler and the condenser.

4.6.5.4 Application to Atmospheric Distillation of Crude Oil

The initial crude oil separation consists in separating the feed into a number
of light cuts in a well-balanced quantity (even though distribution varies from
one crude to another), and one residue, which represents approximately 30 to
50%by weight for conventional crudes. Each of these separations presents the
same difficulty, and this eliminates the criteria mentioned earlier about rela-
tive volatilities. Referring to the preceding sections, particularly to the Tedder
and Rudd diagrams, it can be seen that the best structure is one where the
products are extracted in the order of their boiling temperatures. This pro-
cedure was used in the beginnings of the oil industry. To minimize initial
investments, the inverse structure was proposed. As shown by Liebmann and
Dhole (1995) present-day technology using side stripping and liquid sidedraw
corresponds to a structure where heavy products are removed first. The
advantage is that it allows the structure of the unit to be simplified by group-
ing all the separators in a main column and the energy supply to be located
in one single furnace situated upstream from the main column. To keep all
the high-level calories in it from being recovered entirely in the overhead
condenser, circulating refluxes serve as intermediate condensers (Figs. 4.52
and 4.53). However, this technology is not the most efficient from an energy
standpoint.
The preceding authors have suggested an alternative technology in order
to group all the devices in one single shell while at the same time utilizing the
direct flow diagram structure. The drawback is that it requires two furnaces
and more condensers. Devos et al. (1987) have proposed another type of set
up, the progressive separation scheme, which does not attempt to integrate all
the columns. The advantage is that it needs only one furnace (replacing the
topping furnace) or two furnaces depending on the type of crude or the
desired flexibility.
 Chapter 4. DISTILLATION,
ABSORPTION
AND STRIPPING 189

I
I No circulating
I
I reflux
I
I
I

5 Overflash

Internal reflux flow

4
ResidueJ

Variations in internal reflux in a column equipped with a circulating reflux

(Source: Nelson).

4.7 Batch Distillation

Batch distillation is often used in the chemicals industry when the amounts to
be processed are not large enough to justify a continuous process. It is also
applied in circumstances where certain impurities need to be removed from a
product. The basic structure of a batch distillation column is shown in
Figure 4.54. The boiler provides the heat required to initiate the desired vapor
flow rate. The vapor is dealt with in the upper part of the column where it con-
tacts a countercurrent liquid stream. A batch distillation column has a rectify-
ing zone only, which purifies the light component that is being processed. The
composition in the reboiler becomes a consequence of the policy applied to
the column, it can not be directly kept under control. Additionally, since the
liquid in the boiler gets constantly heavier during processing, the column
remains in transient conditions. The importance of retention vessels will be
discussed later on.
 -
190 Chapter 4. DISTILLATION,
ABSORPTION
AND STRIPPING

Steam

n-32(= LJ
r
Light gas oil
distillation of
crude oil.

Atmospheric residue
b

The product to be processed is let into the column’s reboiler and after a
given time required to get it in thermal equilibrium (generally with total
reflux), the distillate is drawn off at the top and sent to a vessel. As time
elapses the different substances in the feed will be removed one after another
in boiling temperature order and sent to different vessels. Since the objective
of separation is to reach the desired degree of purity at the end of each of the
phases, operating conditions must be optimized over the whole time period to
minimize operating costs. Among other factors, good management of the
reflux ratio allows quality specifications to be met while at the same time com-
plying with recovery ratio constraints and minimizing slop or intermediate
cuts. Between products there are in fact usually off-specification cuts that cor-
respond to lost profits and may advantageously be recycled.

4.7.1 Principal Balances in a Batch Distillation Column

Let us consider at an initial time a quantity of matter Bo with a composition
(z,,~,z ~ ,...,
~ zo,J.
, At a time t, a certain quantity of matter has been drawn off.
 Chapter 4. ABSORPTION
DISTILLATION, AND STRIPPING 191

Basic structure of
a batch distillation
column.

It is characterized cumulatively by a quantity D and a composition ( x ~ xD,2,

,~,
x ~ ,...,
~xD,J.
, The overall balance between these two times is therefore defined
by Eqs. 4.56 and 4.57:
Bo = B, + D (4.56)
zO,i = + DxD,i (4.57)
or:

This same balance is written differentially between two times t and t + dt as

follows:
-dB=dD
- d (Bz;)= d (DxD,;)
or:

(4.58)

(4.59)

If a model can be found to relate the distillate composition to the boiler

composition for a specified number of theoretical trays and reflux ratio, it is
easy to predict the behavior of a batch distillation column.
192 Chapter 4. DlSTILLAT/ON, ABSORPTION AND STRIPPING

4.7.2 Applying the McCabe and Thiele Method

Let us look again at Eq. 4.48, the mass balance of a tray in a dynamic or
unsteady state:

Let us add the hypothesis that the influence of retention terms is negligible
compared to the other terms in Eq. 4.48 to the McCabe and Thiele method
hypotheses discussed earlier (see Section 4.2.3.2). The equation is then
reduced to one of the steady state operation type and the column will be
described by a series of steady states. All the calculation methods previously
discussed are applicable, among them the McCabe and Thiele method for
binary mixtures complying with the Lewis hypotheses.
At the initial time, the boiler has a composition t and the three theoretical
stages allow the composition x, to be reached (unbroken line on Figure 4.55a)
for the chosen reflux ratio. At time t, the composition in the boiler has become
t',which leads to the composition xb (dotted line in Figure 4.55a) for the same
reflux ratio. It is also possible to offset the decrease in boiler composition by
increasing the reflux ratio in order to increase the separating power of each
tray and maintain distillate quality (see Fig. 4.55b). This is true as long as the
reflux ratio is not infinite, which it becomes when boiler and distillate compo-
sition are related by the Fenske law. Distillate quality automatically decreases
from this point onward. Even in total reflux conditions, the light component
mole fraction will decrease with time.
These two cases correspond to two extreme policies that can be imple-
mented for a batch distillation column:
1. Choose a reflux ratio value and keep it constant so that the average com-
position of the recovered distillate meets specifications while the recov-
ery ratio objectives are achieved at the same time.
2. Regulate the reflux flow rate so as to achieve the desired quality at all
times. This approach entails installation of a feedback loop with a con-
centration sensor to be effectively implemented. Additionally, the draw-
back is that the option does not take advantage of the fact that it is
inexpensive to get a very pure product at the beginning of a batch.
The choice between these two options is impossible outside a real context.
A simulator can help find the optimum policy and result in a compromise solu-
tion, i.e. varying the reflux flow rate by stages, which is technologically easy to
implement.

4.7.3 Simplified Design Methods

The basic concepts of continuous distillation (i.e. minimum reflux and total
reflux) can also be applied to batch distillation since these principles involve
cascades of vapor-liquid countercurrent contacting equipment.
 Chapter 4. DISTILLATION,
ABSORPTION
AND STRIPPING 193

Figure 0.8
4.55a
-
0.6
Batch distillation.
Variations in distillate
composition versus time
(constant reflux ratio, 0.4
3 theoretical stages).

0.2

0 0.2 Z' Z 0.6 xb xD 1

&
Figure
4.55b
J

Batch distillation.
Increase in reflux ratio
to maintain constant
distillate quality
(3 theoretical stages). I 1 I ! I
194 Chapter 4. DISTILLATION,
ABSORPTION
AND STRIPPING

At a given time, the boiler has a composition x, and the distillate a compo-
sition x,. The extreme distillation conditions are then defined in the following
way:
1. Total reflux: the minimum number of trays is given by the Fenske equa-
tion relating the two compositions (Eq. 4.17):
XD =aN-
- XB
l-xD l-x, (4.17)

where N is the number of theoretical stages involved in the separation.

This equation shows that N is a decreasing function of x,.
2. Minimum reflux: the minimum reflux is given by the slope of the operat-
ing line relating the distillate’s representative point to point B indicating
the composition of the vapor phase in equilibrium with the boiler liquid.
The minimum reflux increases when the mole fraction of the light com-
ponent decreases.
During operation, the distillation column strips the boiler feed of light com-
ponents, then separation becomes more difficult. Figure 4.56 shows the varia-
tion in the minimum number of stages versus the recovery ratio for the light
component. It also shows that a batch distillation column must be designed
on the basis of the final stripping conditions. When the mixture is multicom-
ponent, the most constraining separation must be sought to ensure proper

R 35
-
?! !
Z
0
30-

5 25-
s= 20-

.-
s
E 15
,,
/
c
iz _ _ _ _ _ _ _ _ _ ----_---
/

10:

5-
I I I I

-
Figure
4.56 Batch distillation. Variation in the minimum number of theoretical trays versus
the required recovery ratio. Initial mixture: 50/50. Specification:98%.
 Chapter 4. DISTILLATION,
ABSORPTION
AND STRIPPING 195

operation for the whole batch. If we think in terms of fixed product composi-
tion, the choice of the number of trays has a dual consequence:
1. It determines the unit’s energy consumption.
2. It influences the recovery ratio, a drop in this rate for multicomponent
mixtures causes the number of slop cuts to increase and they are a
source of lost profits or added costs due to recycling.
All the other simplified methods presented in the section on continuous
distillation are applicable provided the calculation is applied many times so
that the whole distillation batch is represented. Let us insist once again, how-
ever, on the fact that it is only an approximation, which is even more justified
as process retentions are small.

4.7.4 Rigorous Calculation Methods

The dynamic simulation tools presented in the section on continuous distilla-
tion can be applied to this problem. They allow the influence of the different
design parameters, among which the retentions in the condensate drum and
the gas-liquid contacting equipment, to be quantified precisely. The following
example, obtained using PROSIM Batch software, shows the dynamic behavior
of a batch column with 14 theoretical trays with a 5 liter retention, a condenser
with a 50 liter retention and a reboiler. It processes a feed including methanol,
3-pentanone and 2-hexanone according to the following protocol:

Initial Feed:
800 kg of methanol (A), 600 kg of 3-pentanone (B), 600 kg of 2-hexanone (C).
Operating Procedure:

Reboiler power End criterion

Reflux ratio
(kw) (mass fraction)
140 A < 95% in receptacle
127 B > 90% in distillate
%Pentanone 117 B < 95% in receptacle
Slop cut 105 C > 80% in distillate
2-Hexanone 93 Reboiler load < 5 kg

Figure 4.57A shows the reference curves. Figure 4.57B presents the influ-
ence of the reflux ratio for an identical heating protocol. It can of course be
noted that the duration of the batch is shorter since the decrease in reflux
ratio means an increase in distillate flow rate. The curves are a little slower,
which would suggest a loss in terms of recovery ratio at least for the interme-
diate product and the heaviest product. Figure 4.57C shows the influence of
retentions. Methanol’s much slower exit will be noted and although the inter-
mediate substance exits more rapidly, the end of the phase seems to be satis-
factory. The balance for the heaviest component is naturally the resultant.
 A

0.5

0.5

"
1 2 3 4 5 6 l i m e (h)

0.5

Figure
4.57 Batch distillation.
Example of separation between methanol, 3-pentanone and 2-hexanone.
 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING 197

Table 4.12 gives a synthesis of these results expressed in weight for all the
components. It shows that generally speaking the products are of better qual-
ity when the reflux ratio increases (which comes as no surprise) and when
retentions are smaller. This example underscores one of the basic aspects of
batch distillation: in order to get satisfactory separation performance, reten-
tions must be minimized throughout the process. This is why packing is more

c
advantageous than trays.

2-Hexanone cut
-
MeOH 3-P 2-H MeOH 3-P 2-H MeOH 3-P 2-H
-
Reflux ratio = 2
Retention = 5 liters 791.9 41.7 0.004 3.8 351.7 14.7 328.7
-
Reflux ratio = 2
Retention = 50 liters 782.5 41.2 0.003 4.2 459.9 20.0 3.6
Reflux ratio = 1.8
Retention = 5 liters 791.1 41.6 0.004 3.8 308.1 12.4 20.2 311.1

1412 Variations in performance: example o f an M-P-H ternary mixture.

The methods normally used for continuous distillation columns can be

applied without restrictions in calculating gas-liquid contacting equipment.
During operation, the reboiler power and reflux ratio must be chosen so as to
comply with the hydrodynamic operation domain.

4.8 Absorption. Stripping

4.8.1 Absorption
When the target compounds are diluted in a large quantity of gas, in particular
in noncondensable gases, distillation is not a very advantageous operation
since producing a reflux is extremely costly. In addition, the recovery ratio of
the compounds remains limited by the vapor-liquid equilibrium located in the
partial condenser. Here the use of a heavier solvent becomes pertinent. The
principle of absorption consists in placing a rich gas in countercurrent contact
with an absorption liquid that is very poor in added-value components (Fig.
4.58). The high-value components condense in the liquid, thereby causing heat
to be released. Absorption efficiency increases as: the temperature of the
absorption liquid is lower, the operating pressure higher, the absorbent flow
rate greater and the absorbent’s molar mass is lower. Molar mass has limits,
however, since avapor-liquid equilibrium occurs on the first tray. If the solvent
198 Chapter 4. DISTILLATION,
AEISORPTION
AND STRIPPING

is too light, significant losses will take place by evaporation and limit the prof-
itability of the operation. Sometimes a second absorption zone is incorporated
above the first to reduce evaporation losses: an oil (called sponge oil) heavier
than the first one recovers the lost solvent.

r 1- I
f /( I
+ I / I I
) L Y tJ', I
:;,u 1' I
I -_--/ I I
I 1 I
I I I
'7' I
I I
Y I
P' Condensed gas
I
I
I
I
" I

- Stripping fluid I
Unprocessed I G ; %,N+I (steam) I
I
gas I I
I
A
I
I
I
I
I

-
Figure
4.58 Simplified flow diagram o f an absorption unit with regeneration of the absorp
tion liquid by stripping.
 Chapter 4. ~IS'TILLATION,ABSORPTION AND STRIPPING 199

In this equation, Ai is the absorption factor, G and L are the molar flow rates
of gas and liquid respectively. As the concentration in high-value products in
the gas is very small, it can be assumed in an initial approximation that L and
C are constant all throughout the absorption column. The same is true of the
K values, due to small temperature differences from tray to tray in addition to
the reasons mentioned above. Applying Eq. 4.60 to all the trays leads to the
expression:

(4.61)

where:
yi,N+lis the mole fraction of component i in the gas at the absorber inlet
yi,l is the mole fraction of component i in the treated gas (at the exit)
yTo is the mole fraction of component i in the vapor that would be in equi-
librium with the absorption fluid whose composition is xi,o; is close
to
to 0 and therefore y = 0 in most cases
GYip+N+1-G~i.1
represents the absorption rate, usually similar to
GYi,N+I - GY z o
GY;J/+1- GYi.1

Figure 4.59 shows the recovery ratio of component i versus the absorption
factor and the number of theoretical trays involved. It can be noted that there
is practically no more gain beyond some twelve trays.

P
.p
UJ
.-
c
c
2 0.5
0

"
0 0.5 1 1.5 2
. . " Sfactor
Absomtion A or strimina
-
Figure
4.59 Recovery ratio by absorption or stripping uersus A or S.
200 Chapter 4. DISTILLATION,
ABSORPTION
AND STRIPPING

From a process standpoint, two characteristics should be pointed out:

Gas and liquid traffic is set by criteria outside the unit operation. The
unit’s performance is a resultant, there is no degree of freedom whereby
the quality of the treated gas and of the enriched liquid can be adjusted.
When the absorption unit is adiabatic, the heat of condensation is evac-
uated by the enriched oil. However, a temperature profile is created in
the absorber with a maximum in the zone where condensation is at a
maximum (Fig. 4.60). This configuration is detrimental in terms of sizing
because it causes partial revaporization of the products absorbed in the
lower zone in the column. To prevent this internal recycling, heat
exchangers are incorporated in order to approach isothermal operation.
One exchanger every three trays is usually recommended.

Absorbertop I

Absorber bottom .
II - Temperature (“C) II
’1Figure I I

I 1
4.60 Typical temperature profile in an absorber.

4.8.2 Stripping
The enriched oil coming from the absorption unit can be treated in a distilla-
tion or in a stripping column. In this sort of operation, the liquid stream is con-
tacted counter current by a superheated steam stream, thereby causing
revaporization of light components. Steam stripping, which is used later on to
condense the vapors recovered at the top of the column, in fact employs two
phenomena:
The injected steam can be considered in this case as a noncondensable
gas which lowers the partial pressure in the column and makes the
absorbed products vaporize more readily.
 Chapter 4. DISTILLATION, ABSORPTION
AND STRIPPING 201

The steam provides the calories, in the form of sensible heat, that are
required to heat the oil and thereby make degassing the oil easier.
Steam stripping can also be used to revaporize the light components in a
complex mixture. Applications are common in crude oil distillation, where side
strippers top the products that are drawn off (see Chapter 5).
In the same way as for absorption, operating conditions for the unit are
fixed by outside flow rates (rich oil flow rate and steam flow rate). No degree
of freedom exists to adjust the quality of the regenerated oil. The efficiency of
the stripper will be increased when the steam flow rate is increased or when
the operating pressure is lowered. The pressure will, however, have to remain
higher than a minimal value so that overhead vapors can be condensed.
From a calculation standpoint, the numerical techniques discussed in
Section 4.4 can of course be used. It should, however, be noted that care is
required in initializing temperature profiles because product revaporization
creates an inverse temperature gradient compared to the conventional gradi-
ent found in a distillation column.
Simplified methods are also available. The tray by tray mass balance yields
Eqs. 4.62 and 4.63:

(4.62)

(4.63)

If the stripping fluid does not contain any component i, and this is usually
the case, L ) ; , ~ , +=~ 0 and therefore x ? ~ , =+ 0.~
In these equations, Sirepresents the stripping factor, already mentioned in
distillation, and xi,,and xisvare the mole fractions of the component in the rich
and in the poor oil.
Lx;,,,- LX;,N’ Lx;,,,- L X ; , K
defines the stripping rate, usually similar to
LX;,,, - Lx?”+l LX;,,,

The similarity of Eqs. 4.61 and 4.63 means that the same graph defined in
Figure 4.59 can be used.

4.9 Extractive and Azeotropic Distillation

When the relative volatilities of the key components in a distillation separation
are small, distillation becomes extremely costly, from the standpoint of both
investments and operating costs. This is when it is attractive to modify the
physicochemical properties of the mixture by adding a solvent with a particu-
lar chemical affinity to one of the components. The objective is to increase
volatility artificially, which can in some cases cause an inversion of volatility
202 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING

properties. This is the principle of extractive and azeotropic distillation. The

modifications in vapor-liquid equilibria can be obtained by acting either on the
activity coefficients in the liquid phase or on the fugacities in the vapor phase.
In the continuation, we apply the concept of extractive distillation to the
separation of regular mixtures where a third substance modifies the relative
volatilities. Azeotropic distillation concerns the methods that are useful in s e p
arating mixtures where there is at least one azeotrope.

4.9.1 Extractive Distillation

The fundamental problems to be solved in applying this technology are:
searching for the solvent and modeling the mixture that is formed with it.
Although the physicochemical criteria are clear, it is necessary to add other
considerations in the selection procedure. In addition to the aspects of prod-
uct price, corrosivity, thermosensitivity and safety, it is necessary to take the
strength of the bond between the solvent and the extracted molecules into
account. The stronger the bond, the more expensive solvent regeneration will
be. This situation can lead to higher energy consumption for regeneration than
the gain made in the first step. This is one of the reasons why the domain of
application of extractive distillation is limited. It becomes much more advan-
tageous when a whole family of substances can be extracted from a feed.
In addition to all the basic distillation concepts, all the calculation methods
for distillation columns can be applied and used in sizing and optimizing facil-
ities. What remains to be found is the best way to inject the solvent into the
column. Though it is possible to inject the solvent into the feed, it is usually
preferable to inject into the rectifying zone if the ”extracted” substance is due
to end up in the residue in order to shut out the products that would have a
tendency to go toward the distillate. Meanwhile, injection into the stripping
zone is preferred if the “extracted” substance is to end up in the distillate.
When solvents modify the activity coefficients in the liquid phase (see
Chapter 2), they can be compared by means of the infinite dilution activity
coefficient, which defines the extracting power, while the ratio of these param-
eters for different components gives the selectivity.
Tables 4.13 to 4.16 list a number of solvents that can be used in extractive
distillation.

4.9.2 Azeotropic Distillation

In the previous sections we had always assumed that bringing about a vapor-
liquid equilibrium caused the vapor to be enriched in the light component. For
certain binary mixtures, called azeotropic mixtures, there is a composition for
which the vapor phase composition is identical to that of the liquid phase. The
mixture behaves on and off like a substance with a specified boiling tempera-
ture. When the boiling temperature is lower than that of the mixture compo-
nents, the azeotrope is of the minimum type. In the opposite case we have a
 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING 203

Furfural 2-Ethylhexanol
Acetonylacetone Propyleneglycol
Nitrobenzene Hexyleneglycol
Nitrotoluene 2-Ethylhexylamine
Phenol o-Chloramine
Aniline o-Phenetidine
Dichloroethylether o-Chlorophenol
Phenyl cellosolve o-Nitrophenol
Phenol-cresol (60-40) o-Phenylphenol
Cresol Methylsalicylate
Dimethylsulfolane-water Nonanoic acid
Dimethylalinine-aniline Sulfolane
Phthalic anhydride Diphenyloxide

1 4.13 I Solvents that can be used in separating aromatics-paraffins (Dunn et al. 1945).

Selectivity at 25°C
Solvent
YE5/YE5-
N-met hylpyrrolidone 1.96
Acetonylacetone 1.87
Dimethylsulfolane 1.95
Dimethylcyanamide 1.96
Dimethylformamide 1.96
Furfural 1.87
Acetonitrile 2.77
n-Butyrolactone 2.17
bis-Chloropropionitrile 2.18
Pyrrolidone 1.99
Propylenecarbonate 2.10
Nitromethane 2.49
Ethylenediamine 2.11
Ethylenecarbonate (36.8"C) 1.92
I
Table
Solvents that are selective in separating pentenes-pentanes (f15/f15- > 1.87)
(Gerster et a!., 1960).

maximum-type azeotrope. The composition where the phenomenon occurs is

the azeotropic point (see also Chapter 2). Figures 4.61 and 4.62 show dew
point and bubble point curves for such mixtures along with the diagrams relat-
ing the composition of the vapor phase to that of the liquid. When this type of
204 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING

Selectivity at 25°C
Solvent 1
I Solvent 2
Y E5 IY E5-
Methylcellosolve Water (5%) 1.69
Tetramethylenesulfone (5%) 1.70
Tetramethylenesulfone (20%) 1.81
Nitromethane 1.70
F'yridine Water (10%) 1.70
Oxydipropionitrile (1 0%) 1.66
Butyrolactone (32.1%) 1.79
Benzene Tetramethylsulfone (50%) 1.73
Oxydipropionitrile (50%) 1.75
Methylethylcetone Butyrolactone (50%) 1.79

I 4.15 I Mixtures of solvents that are selective in separating pentenes-pentanes

I 1 (Center et al., 1960).

Solvent vol/vol hydrocarbon Selectivity at 25°C

YZ4/YE4-
Nitromethane 1.8 1.60
N-formylmorpholine 4.6 1.60
Aniline 3 1.65
Furfural + 4% water 3.7 1.78
Aniline + 3.5% water 4.4 1.77
Methylacetoacetate + 10%water 3.6 1.67
Phenol + 10%water 2.5 1.66
--
Table
4.16 Solvents that are selective in separating butenes-butanes (E4/flK
> 1.60)
(Hess et al., 1952).

mixture is separated by distillation, it is impossible to go beyond the

azeotropic point. The products of a column are one of the pure components
and the azeotrope. In the example in Figure 4.63, a column separates the feed
F into A and Az. To achieve total separation, techniques adapted to the physic-
ochemical behavior of the mixture must be implemented. Before entering into
these details, let us not forget that calculation methods for distillation
columns remain valid. The difficulty consists in obtaining the thermodynamic
model capable of representing the equilibrium curves in a satisfactory manner.
Methods such as NRTL, Wilson, UNIQUAC and UNIFAC (see Chapter 2) have by
their very nature structures adapted to such calculation, but are limited in rep-
resenting the domain of moderate pressures (lower than a few bars).
 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING 205

‘t

A XAZ B
I

I
I
Minimum-type
azeotrope. L

0 1
X

4.9.2.1 Pressure Sensitive Azeotropes

The azeotropic point generally varies with operating pressure. The principle of
the separation flow diagram consists in using two columns running at different
pressures in order to bypass the azeotrope. Figure 4.64 shows an example of
minimum-type azeotrope that is sensitive to pressure. Let us consider a feed
point F located in the left-hand part of the phase diagram corresponding to the
high pressure Pz. Separation with a distillation column will provide the heavi-
est component B in the residue and the azeotrope Az, in the distillate. The
206 Chapter 4 DISTILLATION, ABSORPTION
AND STRIPPING

TAZ

1 1
9
1

I Figure
4.62 Maximum-type
L I // L _I azeotrope.

0 XAZ x 1

azeotrope then becomes the feed for the second column which operates at a
lower pressure PI. The second column yields the light product A in the residue
and the azeotrope Az, in the distillate, Az, is then recycled t o the first column.
The route illustrated by this specific case can be adapted very readily to
other feed compositions or t o maximum-type azeotropic mixtures. In the sec-
ond case the mixture’s basic components will be recovered as distillate with
the azeotrope a s residue.
 Chapter 4 DISTILLATION,
ABSORPTION
AND STRIPPING 207

0 xR xD I x 1
A A2 B
XR
-
Figure
4.63 A column separating a feed into a product and a distillate whose composition
is close to (and lower than) that o f the azeotrope.

4.9.2.2 Separation by a Third Substance

The technology presented above does not apply to most cases and the
problem can be solved by using a third substance, often called a carrier. The
objective is to bypass the azeotropic point by taking advantage of the charac-
teristics of residue or distillation routes.
a. Residue Curves
Let us consider a flash distillation operation as represented in Figure 4.65.The
development of separation in this type of unit is described by the system of
differential Eqs. 4.64:5
- with yi = Ki(T, P, 2,y)xi
dXi = yi - x i (4.64)
de
where 8 is a dimensionless time equal to In B, /B.
Accordingly, for an initial composition x,,it is possible to plot the variation
in composition of the liquid residue x over time. As Widagdo and Seider (1996)
have reminded us, this parameter 8 is equivalent to a normed packing height.
This type of equation therefore describes the operation of a packing-type col-
umn operating with a total reflux and defines the locus of compositions that
can be reached by the distillate when the packing height increases.

5. Eq. 4.64 is the result of utilizing the following material balance:

Bx, = dByi + (B - dB) (xi - dx> for a compound i such that Ki is greater than 1.
208 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING

T
t

1y
F-

L
B

Figure
4.64 Minimum-typeazeotrope. Pressure sensitiviw.
 Chapter 4 DISTILLATION, ABSORPTION AND STRIPPING 209

Flash distillation

I I

Figure 4.66 gives an example of distillation routes or residue curves for a

non-azeotropic ternary mixture. It can be noted that these curves all originate
in the node corresponding to the lightest component and meet at the node of
the heaviest component. They approach the node characteristic of the median
product to a greater or lesser degree. This topology expresses graphically a
concept that is commonly admitted in distillation, i.e. that it is impossible to
recover the intermediate product by one single distillation step.
It can be seen, however, that in certain circumstances the residue curves
approach point B to such an extent that the median component can be recov-
ered with an acceptable quality by a simple sidedraw.
Let us now consider a distillation separation producing a distillate and a
residue in a packing-type column operating with a total reflux. The mass bal-
ance equations described in Section 4.2.3 are written:

By geometric construction in a triangular diagram, the points characteris-

tic of the feed, the distillate and the residue are located on the same straight
line. The ratio of the lengths of segments FD and FR satisfies the lever rule, it
represents the ratio of streams R/D. In addition, the points representing the
distillate and the residue must be located on the same residue curve (see
Fig. 4.66). By making the balance straight line pivot around point F and staying
on the same residue curve, a whole set of separations is rapidly swept. All that
is needed to define the separation field of a given mixture is to identify the two
210 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING

I C R

Figure
Ternary diagram. Distillation routes. Determining possible residues and dis-
tillates. A: the most volatile component. B: the intermediate component.
C: the least volatile component.

limits. The first consists in recovering one of the products ultra-pure, since the
other effluent is determined by mass balance. Even though they are not r e p
resented in Figure 4.67, there are curves that meet these specifications. The
second limit consists in taking the residue curve passing through the feed
point into consideration. It represents all the distillate compositions that
might be obtained by a packing-type column when the curve is followed
toward the light product. By following the curve toward the heavy product, the
potential residues are identified. In this configuration, the yield in distillate or
in residue is zero as per the lever rule. The shaded areas in Figure 4.67 repre-
sent all the distillates and residues that might be obtained by the separation.
This concept will not be utilized in the rest of the present chapter. It is some-
times used during the process design phase to check the feasibility of column
sequencing.
6. The Concepts of Stable Point, Unstable Point and Stationary Point
If we place arrows going in the direction of decreasing temperatures on the
axes of the triangular diagram, we can note the following:
All the arrows converge toward the heavy product node. This point is
mathematically stable, it corresponds to the convergence point of all the
residues.
All the arrows diverge at the light product node. This point is mathemat-
ically unstable and corresponds to the convergence point of the distil-
lates.
 Chapter 4. DISTILLATION, ABSORPTION
AND STRIPPING 211

L
Figure
4.67 1 Ternary diagram. Areas of partitioning between possible distillates and
residues. A: the most volatile component. B: the intermediate component.
C the least volatile component.

The arrows alternate for the median product. This point is mathemati-
cally a saddle. It corresponds to the point that is impossible for a distil-
lation column to reach.
These statements, which seem trivial for a regular mixture, become much
more interesting for more complex mixtures and make it easier to determine
the potential products of a distillation separation.

c. Distillation Routes
When the description formulated by Eq. 4.64 is replaced by the Fenske rela-
tion, the new trajectories obtained in the triangular diagram are called distil-
lation routes. They describe the variation in composition of the trays in a
column operating with a total reflux, when the number of theoretical trays
increases. Their meaning is similar to that of the residue curves and they are
preferentially applied to analyze flow diagrams involving tray technologies.

d. Application to Azeotropic Distillation

Let us first consider the example of a binary azeotropic mixture and a solvent
(or carrier) whose boiling temperature is between those of the two initial
components. The hierarchy of boiling temperatures is as follows:
TB > Tsolvent > T, > TAz. The rules of the arrows (Fig. 4.68) show that the car-
rier and the light product can not be directly recovered by a simple distillation
212 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING

A (Acetone) A

B Carrier
(Heptane) (Benzene) +B

Separation of a mixture with a minimum-type azeotrope by means of a carrier

(here, benzene).

column. In contrast, if we add a certain amount of solvent to the feed (point ,)'F
the column processing it will manage to produce a residue made up of B pure
and the distillate D, located on the A-solvent axis.
A second column allows component A and the solvent to be recovered.
This twocolumn solution is very ingenious, the solvent is used to move the
feed point away from the AB axis where the azeotrope is located in order to
obtain a distillate on the other axis where there are no obstacles. Once such a
possibility is identified, the solvent ratio remains to be optimized to minimize
operating costs.
Let us now consider the case where the carrier is the lightest of the com-
ponents, this defines the following hierarchy: TB > TA > TAz> Tsolvent. The
arrow rules indicate clearly that the azeotropic point can not be reached.
Depending on the zone where the representative point of the feed is located it
will be possible to recover only one of the components in the residue, while
the distillate will always be a product that is rich in solvent. The zones are
defined by a characteristic residue curve (Fig. 4.69) linking the azeotropic
point directly to the apex corresponding to the solvent. The curve can readily
be identified by integrating the residue equation from the mixture containing
the azeotrope and a small amount of solvent. The curve is critical since it has
been shown that no residue curve can cross this boundary in total reflux con-
ditions. In order to isolate each of the components making up the azeotrope,
it is therefore necessary to find a method in order to move from one side of the
boundary to the other side. When the boundary is strongly curved, a mixture
 Chapter 4. DISTILLATION, ABSORPTION
AND STRIPPING 213

B A
L
Figure
4.69 Ternary diagram with a minimum-type azeotrope between A and B.
Determining possible distillates and residues. Going over the boundary.

such that the representative point is located in the concave part of the curve
can be separated into a distillate and a residue by utilizing a residue curve
from the other zone. This property is fundamental in designing an azeotropic
distillation unit that uses a carrier.
First of all a solvent must be identified giving the boundary curve a curved
shape, then the columns will have to be designed so that one of the streams
has a composition that situates it in the appropriate zone.
Figure 4.70 gives an example of flow diagram design when the feed is rich
in the lightest component. The mixture of the feed with a stream belonging to
the same zone gives the light product in the residue plus a distillate located
near the boundary. A feed belonging to the other zone is obtained by mixing
the distillate with a specified amount of carrier or with a stream rich in carrier.
Distilling this new mixture gives the second component in the residue plus a
distillate near the boundary and included in the convex zone of the boundary
curve. The curve can be crossed by distilling the mixture to get a distillate rich
in carrier recycled to the second column, and a residue rich in light compo-
nent remixed with the feed. This type of set up is much more expensive in
investment than the preceding structure, which only requires two columns.
Triangular diagrams can be used to help select the carrier and design the
basic separation flow diagram. Then what remains is to optimize it based on
rigorous column calculations. At this point, it is important to check whether
the boundary can be crossed. In fact the concept of barrier as we discussed it
214 Chapter 4. DISTILUTION, ABSORPTION AND STRIPPING

B A2 F A

I I
I I
I I
I I
I I
I I
I I
I I
I I
F
t.
i-M:* I
I- I

Figure
4.70 General process for separating a binary mixture with a minimum-type
azeotmpe by means of a carrier with a boiling point lower than those of the
 Chapter 4. DISJILLAJION, ABSORPTION AND STRIPPING 215

earlier is rigorous for columns operating with a total reflux. It is not rigorous
with a finite reflux and this provides an opportunity to simplify the process.
Inversely, it would be costly to diagnose it during the flow diagram design
phase.

e. Selection Criteria for a Carrier

(Stichlmair and Herguijuela, 1993)
The definition of selection criteria for a carrier becomes easier using the prop-
erties highlighted during the analysis of triangular diagrams. Beyond very gen-
eral rules concerning corrosivity, toxicity, etc., two basic rules result from the
analysis:
1. Choose a carrier that does not create a boundary line in the triangular
diagram if possible.
2. When the boundary exists, it must be highly curved to facilitate passing
from one zone to the other.
The preceding principles can be defined more explicitly to provide selec-
tion criteria for carriers:
Processes without Crossing the Boundary
Minimum-type azeotropic mixture: the carrier must either have a boiling
temperature in the same range as those of the feed components, or be
more volatile if it forms a maximum-type azeotrope with the light com-
ponent of the feed. The boiling temperature of this new azeotrope must
be in the same range as those of the initial components.
Maximum-type azeotropic mixture: the carrier must either have a boiling
temperature in the same range as those of the feed components, or be
heavier, provided it forms a maximum-type azeotrope with the heavy
component in the feed, whose boiling temperature is included in the
range of those of the initial components.
Processes with Boundary Crossing
Minimum-type azeotropic mixture: the carrier must have a boiling tem-
perature either:
- lower than all the other boiling temperatures (including the
azeotrope’s),
- in the same range as the feed components, provided it forms a new
minimum-type azeotrope with the light component in the feed,
- or higher than those of the feed components, provided it forms two
new minimum-type azeotropes with the components. At least one of
them must have a boiling temperature lower than that of the initial
azeotrope.
Maximum-type azeotropic mixture: the carrier must have a boiling point
either:
- higher than all the other boiling temperatures (including the
azeotrope),
216 Chapter 4. DISTILLATION,
ABSORPTION
AND STRIPPING

- in the same range as the feed components, provided it forms a new

maximum-type azeotrope with the heavy component in the feed, or
- lower than those of the feed components, provided it forms two new
maximum-type azeotropes with the components. At least one of them
must have a boiling temperature higher than that of the initial
azeotrope.

4.9.3 Heteroateotropic Distillation

The equilibrium diagram for certain mixtures, termed heteroazeotropes, is
represented in Figure 4.71. For a temperature higher than the azeotropic tem-
perature, conventional vapor-liquid equilibria can occur. However, at the
azeotropic point, the liquid phase undergoes demixing into two liquid phases
with each one having a definite composition. Below this temperature, all that
remains is a liquid-liquid equilibrium giving compositions that depend on tem-
perature alone.
Such mixtures are usually separated as shown in the simplified flow dia-
gram in Figure 4.71. The feed is let into the condensate drum subcooled so as
to obtain a phase rich in the volatile component (1) and a phase rich in the
heavy component (L). Two stripping zones isolate the components.
This simple structure can be applied to binary mixtures. For more complex
mixtures, only one zone in the phase diagram may exhibit demixing proper-
ties. The heteroazeotropic behavior of the column then depends on the dis-
tance traveled along the distillation route. With slight modifications in
operating conditions, the location of trays operating in a liquid-liquid-vapor
equilibrium may accordingly change suddenly. Even though present-day simu-
lation tools allow this type of calculation to be made (provided the thermody-
namic model is known), designing and operating such processes are extremely
difficult.

4.10 Reactive Distillation

The term "reactive distillation" is used for processes that deliberately associ-
ate a distillation separation operation and a chemical reaction. In the batch
process version, a rectifying zone is incorporated above the reactor in order
to take off one of the products. In continuous operation it is sometimes nec-
essary to introduce a catalytic zone between two gas-liquid contacting equip
ment zones.
In any case, chemical reactions very often occur in the liquid or vapor
phases inside distillation columns. Some of the most frequent examples are
the polymerization and thermal cracking reactions in the high temperature
zone in the column. Such phenomena can be modelled by adding a description
of the reaction in the descriptive equations for the trays, which become:
Total mass balance:
Fp + LPl + V,+l = (V, + S:) + (L, + Sb) (4.65)
 Chapter 4. DISTILLATION,
ABSORPTION
AND STRIPPING 217

I I
I
I I

A XAZ B

1
Figure
4.7 1 Heteroazeoh-opic distillation.
218 Chapter 4. ~lSTILLATlON,ABSORPTION AND STRlPPlNG

Partial mass balance:

Fpz; + L p - 1 ~ i , p - l + V p + l Y i , p + l + rig= ( V p + S z ) ~ i , p+ ( L p + SpL)xi,p (4.66)
Enthalpy balance:
i=n

~ p h F , p+ ~ p - 1h k l + ~ p + hl : + l + C
i = 1 hi,forrnation ri.p

= (Vp + S z ) h: + (Lp + Sp")hp" (4.67)

where ri,p(mol/s) represents the rate at which component i is consumed or
produced on tray p. An equation defining rip with respect to the composition
of the liquid phase, the residence time on the tray, the pressure and the tem-
perature must be available.
In Volume 3, Chapter 2, examples are developed of calculating heat of reac-
tion from the components' formation enthalpy, and expression of kinetic laws.
Most software can solve these basic equations, but it should be noted that
the presence of a chemical reaction often causes convergence problems
related to the presence of a concentration pinch when operating conditions
are not properly chosen.

4.10.1 Batch Coupling of a Reactor and a Rectifying Zone

Let us assume that a balanced reaction of the type A + B H C + D takes place
in the reactor and that one of the products is the lightest component in the
mixture. Incorporating a rectifying zone above the reactor can eliminate one of
the products little by little and move the equilibrium over completely.
In addition to the criteria mentioned before, the reaction conditions must
of course be compatible with the vapor-liquid equilibrium conditions.

4.10.2 Catalytic Distillation

Catalytic distillation is a process whereby a zone in the column is filled with a
catalyst where the reaction takes place. Depending on the configuration, the
catalyst can be in a zone where vapor-liquid equilibria take place at the same
time or it may work only on one of the two phases.
The major advantage of catalytic distillation is that it reduces investments.
In a conventional process where reaction conversion is limited by chemical
equilibrium, the reactor effluent is sent to a separation section in order to
recycle the products that have not yet reacted. This configuration means high
investments. This tendency is amplified for exothermic reactions, because the
use of an adiabatic reactor causes a rise in temperature in the reaction zone,
which is detrimental to equilibrium, and an increase in the recycle flow rate.
This effect of temperature can be limited by heat exchangers but they are
expensive in investment costs.
Catalytic distillation is consequently applied preferentially to equilibrium
exothermic reactions. Internal recycling in the column, which is also used to
 Chapter 4. DISTILLATION, ABSORPTION AND STRIPPING 219

separate products, can enhance conversion. The presence of liquid offsets

exothermicity, part of the liquid is vaporized, thereby maintaining the thermal
level.
In the refining industry, catalytic distillation is used to produce ethers
(MTBE, ETBE) (see Vol. 3). This application is one of the rare cases for the
industry where distillation conditions are compatible with reaction conditions
and the catalyst is capable of withstanding the specific conditions of use.