Energy 37 (2012) 417e429

Contents lists available at SciVerse ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

A real column design exergy optimization of a cryogenic air separation unit

J. Rizk*, M. Nemer, D. Clodic
Mines ParisTech e Center for Energy and Processes, 5 rue Lon Blum, 91120 Palaiseau, France

a r t i c l e i n f o a b s t r a c t

Article history: Distillation columns are one of the main methods used for separating air components. Their inconve-
Received 10 March 2011 nient is their high energy consumption. The distillation process is simulated in three types of columns
Received in revised form and the exergy losses in the different parts calculated. A sensitivity analysis is realized in order to
4 August 2011
optimize the geometric and the operational parameters of each type of column. A comparative exergy
Accepted 6 November 2011
analysis between the distillation columns considered for cryogenic air separation shows that the exergy
Available online 12 December 2011
efciency of a double diabatic column, with heat transfer all through the length of the column, is 23%
higher than that of the conventional adiabatic double columns.
Keywords:
Exergy analysis
2011 Elsevier Ltd. All rights reserved.
Distillation columns
Diabatic process
Cryogenic air separation

1. Introduction nitrogen-oxygen mixtures, higher energy consumption is needed

Three processes for air separation exist: cryogenic distillation,
pressure swing adsorption, and membrane separation. The choice
of the process to be used is based on the desired products. Cryo-
genic air distillation is used when product high purity is needed. It
is also advantageous when products are required in liquid form.
The pressure swing adsorption process is used for high nitrogen
purity (98e99.5 volume percentages). The membrane separation
process is used for low ows and low purities (less than 98
volume percentage). In this article, we consider the cryogenic
distillation process, which is the most widely used process for air
separation [1].
In order to separate the air components by a distillation process,
a feed stream, containing nitrogen, oxygen, and argon is sent into
a distillation column, operating between a heat source (boiler at the
bottom) and a cold source (partial condenser at the top), and
generating mass transfers between a vapor stream owing upward
and a liquid stream owing downward. This type of column, the
simple adiabatic column (Fig. 1), constitutes the conventional
distillation column, which is the reference for bench marking new
distillation concepts.
The conventional distillation process is known to be a major
source of energy consumption in the chemical industry. Besides,
when separating mixtures with low relative volatilities, such as
* Corresponding author. Tel.: 33 169194247; fax: 33 169194501.
E-mail address: joelle.rizk@mines-paristech.fr (J. Rizk).
[2]. In order to reduce the energy consumption of distillation
columns, many new techniques have been proposed since 1950 [3].
These techniques include essentially thermal coupling between
two simple columns, or heat-integrated columns.
Thermal coupling between columns consists in two counter-
current streams, a liquid stream and a vapor stream, connecting the
two columns and exchanging heat between them. Several cong-
urations of thermal coupling between columns are studied and
optimal congurations searched for different cases of feed
mixtures.
Heat-integrated distillation process consists in adding or
removing heat from streams all along their ow in the column.
Flows are in consequence diabatic, and the duties on the top and
bottom heat exchangers are reduced, or even totally removed.
In this paper, three congurations of cryogenic distillation
columns for air separation are studied: a simple adiabatic column,
two thermally coupled adiabatic columns, and a diabatic column.
In order to compare the three columns, two different
approaches could be considered: an energy analysis based on the
rst law of thermodynamics and an exergy analysis based on the
second law of thermodynamics [4]. An exergy analysis is necessary
when various forms of energy are used in a process, because it
describes the quality of energy, while energy analysis describes
only quantitative aspects.
In distillation columns, thermal exergy, at the heat exchanger
levels, is converted into chemical exergy, during the component
separation of the mixture [5]. An exergy analysis will provide
information on the quality of the energy involved in the process

0360-5442/$ e see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2011.11.012
418
Fig. 1. Simple adiabatic column for cryogenic air distillation.
with the location and magnitude of the local exergy losses [4].
Besides, it has been proven that, a distillation process with
minimum energy consumption is not necessarily the process with
the highest thermodynamic efciency [6].
An exergy analysis is therefore conducted in order to nd the
thermodynamic optimum for each conguration considered in this
study, dened by its geometric and operating parameters, and to
compare their thermodynamic efciencies.
The thermodynamic efciency of a distillation column is the
ratio of the minimum work required for performing the separa-
tion to the effective work required for the separation [6]. The
effective work of separation is the sum of the minimum required
work and the exergy losses in the system. In order to maximize
the thermodynamic efciency of distillation schemes, exergy
losses are to be minimized, by reducing or eliminating any source
of irreversibility in the process. Irreversibility in a conventional
distillation scheme occurs in the heat exchangers due to the
difference of temperatures between streams, and all through the
length of the column due essentially to mass transfers between
liquid and vapor ows [7]. Thermally coupled and heat-integrated
columns present the advantage of reducing the exergy losses and
thus increasing the thermodynamic efciency of the distillation
process.
2. Simple adiabatic column
Fig. 1 shows the conventional adiabatic column for air separa-
tion. Air in vapor phase is introduced at a feed location F. A nitrogen
rich vapor product is collected at the top and an oxygen rich liquid
product at the bottom of the column. A heat ow Wboil is supplied
to the boiler, where a fraction B of the liquid leaving the stage 1 of
the column is evaporated and is sent back to the stage. A heat ow
Wcond is removed from the condenser, where a fraction C of the
vapor leaving stage N of the column is condensed and sent back, in
liquid form, to stage N.
In order to calculate the exergy losses due to mass transfers in
the column, the variations of mass ow rates, compositions,
temperatures, and pressures of the liquid and the vapor streams
have to be known rst. For this aim, the column is rst modeled and
its operating parameters calculated. Then the exergy losses are
calculated.
J. Rizk et al. / Energy 37 (2012) 417e429
2.1. Simulation of a simple adiabatic column
In order to simulate the distillation process in a simple adiabatic
column, an industrial packing (Fig. 2) is considered, with the
following characteristics [8]:
 Packing surface area ap 244 m2/m3.
 Packing porosity 0.98.
 Height of the packing element hpe 0.197 m.
 Corrugation height hP 0.012 m.
 Corrugation base length bP 0.0225 m.
 Corrugation side length sP 0.0165 m.
 Corrugation angle with the horizontal a 45 .
The evolution of operating parameters is obtained using a nite
volume method. The column is divided into N nite volumes, and
the values of the liquid and vapor ow rates, compositions, pres-
sures, and temperatures are calculated by resolving the conserva-
tion equations on each nite volume. In order to simplify the
problem, the feed stream is taken to be a binary mixture of nitrogen
and oxygen, with a molar composition of 78.1% of nitrogen and
21.9% of oxygen.
Fig. 3 represents a nite volume i, with the different corre-
sponding liquid and vapor uxes. On each nite volume, variables
to be calculated are those noted in Fig. 3.
The main physical phenomenon to be modeled in the distilla-
tion column is the mass transfer between the liquid and the vapor.
The nitrogen being the more volatile component, it tends to
evaporate. The vapor is then enriched with nitrogen while it ows
upward (yi > yi1). The oxygen present in the vapor tends to
condensate, and the downward liquid ow is enriched with oxygen
(x0 i > x0 i1).
Masse ow rates and molar fractions of oxygen and nitrogen are
calculated by solving the mass and species conservation equations
(Eqs. (1)e(5)). In order to simplify the resolution, it is possible to
reduce the number of variables, by adding the two species
conservation equations, and using the relation between the mass of
nitrogen transferred to the vapor and the mass of oxygen trans-
ferred to the liquid (Eq. (6)).
Mass conservation:
Gi1  Gi Li1  Li 0 (1)
Nitrogen conservation in liquid:
_ N2 0
Li1 xi1  Li xi  m (2)
Nitrogen conservation in vapor:
_ N2 0
Gi1 yi1  Gi yi m (3)
Fig. 2. Basic geometry and dimensions of the packing [5].
 J. Rizk et al. / Energy 37 (2012) 417e429 419

coefcients are those presented in [8] and the equation for the
interfacial surface is the one presented in [10].
The mass transfer coefcient in the vapor kG, is represented as
an average value between a laminar coefcient and a turbulent
coefcient (Eq. (12) [8]). The laminar and turbulent Sherwood
numbers are expressed by the VDI-Wrmeatlas correlations (Eq.
(13)). In Eqs. (12) and (13), ScG is the vapor Schmidt number, (dh/
lG,pe) is the ratio of the hydraulic diameter to the length of vapor
ow channel within a packing element, ReG,rv is the vapor Reynolds
number, calculated with the relative velocity (vG,e vL,e), and g is
the contact angle. The liquid/vapor friction factor xGL, is correlated
using the Colebrook equation for the friction factor in circular
channels, modied to have an explicit expression (Eq. (14) [11]) and
the hydraulic diameter dh (Eq. (15)) is calculated as a function of the
Fig. 3. Liquid and vapor ow rates in a nite volume i.
packing geometric parameters and the liquid lm thickness, given
by Nusselt formula for falling lms on inclined wall situation (Eq.
(16)).
Oxygen conservation in liquid:

Li1 x0i1  Li x0i m

_ O2 0 (4) q q
D
kG k2G;lam k2G;turb G Sh2G;lam Sh2G;turb
dh
s
Oxygen conservation in vapor: (12)
1=3 d
ShG;lam 0:664ScG ReG;rv h
Gi1 y0i1  Gi y0i _ O2 0
m (5) lG;pe
Relation between mass ows of oxygen and nitrogen
transferred: x f " !2=3 #
ReG;rv ScG GL dh
ShGt ;urb 8
r

_ O2
m
hfg;N2
_ N2
m (6) xGL f  1 lG;pe
(13)
2=3
hfg;O2 1 12:7 ScG
8
Adding Eqs. (2) and (3) to Eqs. (4) and (5) respectively, and using Eq.
(6), eliminates the oxygen molar fractions and transferred ow rate    

2
from the variables to be calculated. Eqs. (7) and (8) replace Eqs. (4) d=dh 5:02 d=dh 14:5
and (5) in the system to be solved. xGL  2log  log (14)
3:7 ReG;rv 3:7 ReG;rv
hfg;N2
Li1  Li  mN2 _ N2 0
m (7)
hfg;O2  2
bp hp  2dsp
dh  bh (15)
hfg;N2 2  2 0:5
Gi1  Gi mN2  _ N2 0
m (8) bp hp  2dsp bp hp  2dsp bp hp  2dsp
hfg;O2
2hp bp 2hp
The nitrogen molar ow transferred from the liquid to the vapor
is calculated using Ficks law for mass transfer [9].  1=3
3mL vLs


dL (16)
dNN2 k
 rL gap sin a
kG P y*N2  yN2 L xN2  x*N2 (9)
dSi Vm The mass transfer coefcient in the liquid is correlated by Bravo
On a nite volume i, the mass transfer equation gives the mass et al., based on the penetration theory [8].
of nitrogen transferred m _ N2 as a function of the nitrogen molar
s
fractions in the liquid, the vapor and on the vapor/liquid interfacial
DL vLf
surface. kL 2 (17)
0:9pdh

 k

_ N2 kG Si Pi y*i  yi L Si xi  x*i
m (10) The evaluation of the interfacial surface Si between the liquid
Vm
and the vapor ow should take into consideration the liquid
In Eqs. (9) and (10), x* and y* are the equilibrium molar fractions distribution. It is equal to the wetted surface of the packing,
of nitrogen at the vapor/liquid interfacial surface. They are related calculated as a function of the ow characteristics (the dimen-
by Henrys law (Eq. (11)), Ki being the Henrys constant [9]. sionless numbers) and the packing characteristics. Eq. (18) [10] is
then used to calculated the interfacial surface; the factor FSE
y*i Ki x*i (11) accounts for variations in surface enhancement, and the contact
angle g accounts for the surface wettability. For the packing
Three more parameters should be evaluated in order to calculate considered in this study, with a surface area of 244 m2/m3 and
the mass of nitrogen transferred to the vapor ow: the mass a porosity of 0.98, FSE 0.35 and cos(g) 0.9 [10].
transfer coefcient in the liquid kL, the mass transfer coefcient in
the vapor kG, and the interfacial surface Si. These parameters are 29:12WeL FrL 0:15 s0:359
Si p
correlated in many studies, and similar values are calculated using FSE 0:2 (18)
different correlations. In this study, equations of the mass transfer
ap ReL 0:6 1  0:93cos gsin a0:3
420 J. Rizk et al. / Energy 37 (2012) 417e429

Replacing m_ N2 by its values (Eq. (10)) in Eqs. (2), (3), (7), and (8) v2
gives the four equations (Eqs. (19)e(22)) used in the system to be DP 1:941h0:446
L f rg (23)
dh
solved to calculate the variables of the nite volume.

hL 0:0273Re0:331
L (24)
xi  x*i
Li1 xi1  Li xi  kL;i Si 0 (19) The friction coefcient in Eq. (23) is calculated as a function of
Vi
the vapor Reynolds number, as follows [12]:

 
Gi1 yi1  Gi yi kG;i Si Pi y*i  yi 0 (20) 8:02ReG 105 0:364; if ReG < 1500
f (25)
1:43ReG 106 0:266; if ReG  1500
xi  x*i hfg;N2 xi  x*i It is remarkable that equations for the pressure drop, the mass
Li1  Li  kL;i Si kL;i Si 0 (21) transfer coefcient, and the interfacial surface require the vapor
Vi hfg;O2 Vi
and the liquid thermodynamics properties. These properties being
function of variables to be calculated (molar fractions, pressures,
and temperatures), an iterative procedure is used. It could be

 hfg;N2
* 
Gi1  Gi kG;i Si Pi y*i  yi  kG;i Si Pi yi  y i 0 (22) summarized by the following steps.
hfg;O2
1. Given the feed mass ow and molar composition, assume
When the conservation equations are to be written at the stage F
constant mass ow rates of liquid and vapor and linear varia-
where the feed is introduced to the column, Eqs. (19)e(22) are
tions of the liquid and vapor compositions through the N nite
replaced by Eqs. (19a)e(22a) in order to take into account the ow
volumes.
rates added.
2. Given the pressure at the boiler, assume constant pressure on
all nite volumes.
xF  x*F
LF1 xF1  LF xF  kL;F Si 0 (19a) 3. Using the pressure and the molar fractions from 1 and 2,
VF calculate the thermodynamic properties of ows at the liquid

 saturation temperature, using REFPROP 8.0.
GF1 yF1  GF yF kG;F Si PF y*F  yF m
_ F yF 0 (20a) 4. Using the vapor mass ow rates from 1 and the thermody-
namic properties from 3, evaluate the pressure drop using Eq.
(23).
xF  x*F hfg;N2 xF  x*F 5. Using the new pressure prole from 4 and the molar fractions
LF1  LF  kL;F Si kL;F Si 0 (21a)
VF hfg;O2 VF from 1, calculate the liquid saturation temperatures and new
thermodynamic properties for the mixtures.

 hfg;N2
*  6. Using thermodynamic properties from 5, calculate mass
GF1  GF kG;F Si PF y*F  yF  kG;F Si PF yF  yF transfer coefcients and interfacial surface.
hfg;O2
_F 0
m (22a) At this stage, six variables are to be found for each nite volume (Gi,
The evolution of the pressure in the column is function of the Li, xi, yi, xi*, yi*). The total number of variables is 6N. At the 1st and
pressure drop on each nite volume. Many studies have been the Nth nite volumes, four variables are added: the liquid ow
realized in order to correlate the pressure drop in structured sent to volume N (LN1, xN1) and the vapor ow sent to volume 1
packing [12e14]. They correlate the pressure drop using the linear (G0, y0). Equations needed to calculate these variables are the mass
pressure drop in a dry channel and applying corrections to the and species conservation at the boiler and the condenser:
vapor velocity and the friction factor.
G0 BL1
A correlation for the vapor velocity correction, as a function of
y 0 x1
the liquid and vapor properties and of the liquid velocity is (26)
LN1 CGN
proposed by Woerlee et al. [13]: two types of corrections are xN1 yN
applied to the velocity, a static correction, related to the decrease of
the channel section due to the liquid ow, and a dynamic correc- The iterative procedure continues after step 6 by the following:
tion, related to the interfacial velocity. Fernandez et al. [14] ach-
ieved CFD simulations of countercurrent liquid and vapor ows in 7. Write a (6N 4)  (6N 4) matrix A, with Eqs. (1), (11),
a structured packing, by correcting the vapor ow velocity, stati- (19)e(22) written for each nite volume, and Eq. (26). Write
cally and dynamically. Values of the pressure drop calculated when a (6N 4) vector B, with constant values at the feed location.
applying both corrections on the vapor velocity did not differ much 8. Resolve the system AX it B.
from those obtained when applying the static correction only. 9. Using the mass ow rates and molar fractions from 8, repeat
Using this result, only a static correction is applied to the vapor steps 3e8. Calculate a new vector X(it1).
velocity in this study. 10. Calculate the relative differences between each line of vectors
Two correlations for the correction of the friction factor are X(it) and X(it1).
proposed in the literature in [12,14]. The friction factors 11. If the maximum of the relative differences is less than 104, go
calculated by each correlation are compared, and their values did to step 12. Else, go back to step 9, using the vector X(it1).
not differ signicantly. The correlation proposed by Darcy and 12. Exergy analysis, as shown in Section 2.2.
Weisbagh (Eq. (23)) calculates this factor as a function of the liquid
holdup [12]. The liquid holdup, dened as the fraction of liquid 2.2. Exergy analysis of a simple adiabatic column
occupying an element of vertical pipe, is given in Eq. (24), within
a 15% error, as a function of the liquid Reynolds number (correla- The exergy analysis of a simple adiabatic column accounts
tion of Collodziej). for exergy losses due to mass transfer in each nite volume and
 J. Rizk et al. / Energy 37 (2012) 417e429 421

for exergy losses due to heat transfer in the boiler and the conditions on the liquid product (mass ow rate and purity). Exergy
condenser [6]. losses in the process are only evaluated for the columns providing
On a nite volume i, exergy losses are calculated by an exergy the desired product. These conditions allow the algorithm to
balance between the inlet ows and the outlet ows (Eq. (27) [7]). reduce the list of parameters, as follows:

Exi loss Li1 ExLi1 Gi1 ExG L G

i1  Li Exi  Gi Exi (27) 1. For a combination of H, F, P0, B, and C, initialize Mfeed for 5000
tons per day of the liquid product.
The exergy of a uid is dened as the maximum useful work
2. Estimate the diameter d for reduced total pressure drop.
possible during a process that brings it to its equilibrium state [15]:
3. Solve the equation AX B (by applying all steps listed in the

 previous paragraph).
Ex h  href  Tref s  sref (28) 4. If maximum pressure drop exceeds 500 paper meter, correct d.
5. If (1  B)L1 s 5000 tons per day, correct Mfeed.
The exergy loss in the condenser is:
6. Repeat 3e5 until convergence.
 
 7. If oxygen molar fraction in the liquid at the bottom >95%,
_ cool Excool;in  Excool;out LN1 ExG
Exloss;cond m L
N  ExN1 calculate exergy losses and duties of the boiler and the
(29) condenser. Save the results.

And the exergy loss in the boiler is: The algorithm output data are exergy losses for each combina-

   tion of the above parameters. Before proceeding to the overall
Tref
Exloss;boil G0 ExL1  ExG
0 Wboil 1  (30) optimization of the exergy losses in the simple adiabatic distillation
Tboil column, a sensitivity analysis shows the inuence of each param-
The thermodynamic efciency of the column equals the ratio of eter on the total exergy losses in the column and on the distribution
the minimum work required to the distillation process to the of losses in the different parts of the column.
effective work required (Eq. (31) [16]). The minimum work is
calculated by the column exergy balance (Eq. (32) [17]). 2.2.1. Sensitivity analysis
In order to study the effect of each parameter on the exergy
Wmin losses in the process of cryogenic separation of air components in
h P (31)
Wmin Exloss a simple adiabatic distillation column, each parameter is varied
alone for a set of xed values of the others.
X X Fig. 4(a) shows the effect of the reux ratio at the bottom of the
Wmin Exout  Exin (32)
column, B, on the total exergy losses. Fixed values of the rest of the
Before comparing the simple adiabatic column to thermally parameters are listed below. Unless different values are given, these
coupled and heat-integrated distillation columns, an optimum xed values are the one used all through the sensitivity analysis for
conguration, dened by geometric and operating parameters, is the simple adiabatic distillation column.
searched for the simple conguration. Parameters to be optimized
are as follows:  H 10 m;
 d 18 m;
 Column diameter, d.  F 1 m;
 Feed location from the bottom, F.  P0 110 kPa;
 Pressure at the boiler, P0.  C 0.4.
 Reux ratio at the bottom of the column, B.
 Reux ratio at the top of the column, C. The feed mass ow rate varies with the reux ratio at the
bottom of the column and when B increases the feed mass ow rate
The molar composition of the feed stream is 78.1% of nitrogen increases. This leads to an increase in the energy needed to separate
and 21.9% of oxygen. The desired product is low pressure, oxygen the feed mixture components on one hand, and on another hand,
rich liquid (>95% of oxygen). The mass ow rate of products is that an increase in the pressure drop in the column. Consequently, total
of an actual distillation unit, 5000 tons per day. exergy losses in the column increase. The thermodynamic ef-
Parameters listed above are the input data of the simulation ciency of the column decreases from 29% to 15.2% when varying the
algorithm. Once step 11 is achieved, the algorithm veries reux ratio at the bottom from 0.56 to 0.83. It should be noted that,

Fig. 4. Total exergy losses in the column, (a) as a function of the reux ratio at the boiler stage, (b) as a function of the reux ratio at the condenser stage.
422 J. Rizk et al. / Energy 37 (2012) 417e429

for lower values of B, the purity of 95% of oxygen in the liquid

product could not be reached.
Fig. 4(b) shows the variation of exergy losses with the reux
ratio at the top of the column, for a xed value of B of 0.6 and a 20-
m column diameter. An increase in the reux ratio at the top of the
column causes an increase in the nitrogen rich uxes in the
column; exergy losses during the separation increase.
The pressure inuence on total exergy losses in the distillation
process is showed in Fig. 5, for reux ratios B 0.6 and C 0.3. In
order to explain the decreasing value of exergy losses when the
pressure level increases, the distribution of exergy losses in the
different components of the column (boiler, column, condenser,
and compressor) are calculated for each pressure level. Fig. 6 shows
these exergy losses as a function of the pressure. It is remarkable
that exergy losses in the compressor represent a maximum ratio of
Fig. 6. Distribution of the exergy losses in the column for different pressure levels.
13% from total exergy losses. The highest contribution is that of the
column, presenting a minimum ratio of 50% from total exergy
losses. These ratios explain the decrease of the exergy losses when
the pressure increases: even though the exergy losses in the calculated. This minimum value of the exergy losses is calculated in
compressor increase, their contribution in the total exergy losses is a 20-m diameter, 10-m height column, where the feed ow is
signicantly less than that of the column, where exergy losses introduced at a 1-m height from the bottom. This columns operates
decrease with the pressure level. at 110 kPa with a reux B 0.6 at the boiler and a reux C 0.3 at
The feed location inuence on exergy losses is represented in the condenser.
Fig. 7, for C 0.3 and B 0.6. The total pressure drop in the column Fig. 10 shows the contribution of each component of the distil-
decreases from 5 kPa for a feed at 2 m from the bottom to 3.5 kPa lation scheme considered in total exergy losses. The column has the
for a feed at 5 m from the bottom. Less pressure drop leads to less highest exergy losses (57%), followed by the condenser (29%), then
exergy losses, and an increasing thermodynamic efciency is the boiler (11%), and the compressor (3%).
calculated. It increases from 0.27 to 0.29 when the feed was moved This distribution shows that a possible decrease in total exergy
from 2-m height from the bottom of the column to 5-m height. losses is achieved by reducing exergy losses on heat exchangers and
Fig. 8 represents the variation of exergy losses with the diameter the irreversibility in the distillation column. Reducing or elimi-
of the column with a xed value of the column height of 10 m. A nating heat exchangers leads to the conguration of thermally
larger diameter of the column causes lower pressure losses in coupled distillation columns. Reducing the irreversibility in the
the column and thus lower exergy losses for the same separation column is achieved by making the heat transfers all along the
work. column.

2.2.2. Overall optimization of column

In order to choose the geometric and operating parameters of
the column leading to minimum exergy losses, an overall optimi-
zation (based on a genetic algorithm) of the column is realized. For
each combination of the parameters cited above, total exergy losses
are calculated. A decreasing curve is obtained (Fig. 9) and the
optimal congurations deduced.
Fig. 9 shows that the minimum exergy losses in simple adiabatic
distillation columns producing 5000 ton per day of oxygen rich
low-pressure liquid are close to 30,000 kW. The thermodynamic
efciencies could be deduced and values close to 29% are
3. Thermally coupled distillation column
Because of the high energy consumption and the low thermo-
dynamic efciency of the conventional distillation schemes,
researches are realized to nd methods to improve performances of
distillation processes [18]. Besides, in the exergy analysis of the
simple adiabatic column, the cold source for the condenser at the
top is considered to be free, which is not true in real applications.
The liquid nitrogen necessary for condensation is produced in
a separate process, which exergy losses and performances should
be taken into consideration for the optimization. To include the
production of the cold source for the condenser in the process

Fig. 5. Inuence of the pressure on the total exergy losses. Fig. 7. Inuence of the feed location on total exergy losses.
 J. Rizk et al. / Energy 37 (2012) 417e429
Fig. 8. Inuence of the column diameter on total exergy losses.
study, it is considered produced in a distillation column, at low
pressure, thermally coupled with a higher pressure column.
Thermal coupling between adiabatic conventional columns has
been studied since the early fties. It consists in eliminating the
boiler or/and the condenser in one distillation column and
replacing it by countercurrent connections with the second
column. Fig. 11 shows a fully thermally coupled distillation process
of a ternary mixture, also called a Petlyuk column [19].
Based on the Petlyuk conguration, many alternative schemes
have been proposed to improve the thermodynamic efciencies of
the thermally coupled columns, for ternary mixtures distillation
[6,19e21]. Optimal congurations are searched for different rela-
tive volatilities and feed compositions, and some of the proposed
congurations provide similar or even higher efciencies than
those of the Petlyuk columns. Besides, Jimnez et al. note in their
study [19] that the Petlyuk column, because of its interconnections,
requires that the pressure at the bottom of column 1 (to the left) be
lower than the pressure at the bottom of column 2 (to the right),
while the pressure at the top of column 1 should be higher than the
pressure at the top of column 2.
Attempts have been made to introduce the Petlyuk column
conguration into cryogenic air separation applications [6]. A
thermodynamic efciency of 14.9% was calculated, when the
thermodynamic efciency of a single coupling column reaches
44.5%. Therefore, the Petlyuk conguration has not found applica-
tion in cryogenic air separation.
Thermally coupled columns used for cryogenic air separation
are usually simple coupled congurations [22e24]. They are based
on a low- and a high-pressure distillation column. The condenser of
the high-pressure column HP is cooled by the boiler of the low-
Fig. 9. Total exergy losses in different congurations of simple adiabatic columns.
423
Fig. 10. Contribution of components in total exergy losses.
pressure column LP. Fig. 12 shows a representative scheme of the
coupled column.
In this scheme, the air compressing unit has the largest contri-
bution in the total exergy losses (nearly 20% [1]). The pressure in
the HP column is 620 kPa. An improvement could be realized by
reducing the operating pressure.
In this paper, the thermally coupled column of Fig. 12 is ther-
modynamically optimized. Geometric and operating parameters
leading to the minimum total exergy losses are dened.
3.1. Simulation of a thermally coupled adiabatic column
The simulation algorithm of the simple adiabatic column is
modied and used for the simulation and the exergy analysis of the
double column. The HP column has the same structure as the
simple adiabatic column; it has a boiler at the bottom and
a condenser at the top and the distillation products are an oxygen
rich liquid at the bottom and a nitrogen rich vapor at the top. It
differs from the simple adiabatic column only by the operating
pressure. In the LP column, a liquid air feed is sent to the column. At
the top of the column, no condenser is used. A fraction (1-B1) of the
liquid collected at the bottom of the HP column is expanded and
Fig. 11. Representative scheme of a Petlyuk distillation column [19].
424 J. Rizk et al. / Energy 37 (2012) 417e429
Fig. 12. Thermally coupled column for cryogenic air distillation.
sent at the top of the LP column. Eq. (26) of Section 2.1 are therefore
replaced by Eq. (33) for the low-pressure column resolution:
G0 BL1
in each conguration. As stated in Section 2.2.1, the feed mass ow
rates are subject to variation in order to ensure the energy balance
at the intermediate heat exchanger condition and the 5000 tons/
day liquid product mass ow rate. These two parameters are
therefore function of other parameters listed and will not be
considered in the optimization procedure. The algorithm for
calculating exergy losses in the column for each combination of
parameters includes the following steps:
1. For a given combination of the reux ratios at the top and at the
bottom of columns, estimate initial values for Mfeed,HP and
Mfeed,LP, considering a liquid low-pressure product of
5000 tons/day and the energy balance at the intermediate heat
exchanger.
2. Simulate high-pressure column and then low-pressure column.
3. When convergence is obtained, correct Mfeed,HP and Mfeed,LP,
considering the 5000 tons/day liquid product ow rate and the
energy balance at the intermediate heat exchanger.
4. If the new values of the feed ow rates calculated differ from
previous values (relative difference > 104), go back to step 2.
Else, go to step 5.
5. If the oxygen molar fraction in the low-pressure liquid product
exceeds 95%, calculate the exergy losses in the column. Save the
results.
As for the simple adiabatic column, a sensitivity analysis is
realized in order to study the effect of each parameter on the total
exergy losses in the column. Then the overall optimization of the
performance of the column gives the combination of parameters
leading to minimum exergy losses in the distillation column.
y0 x1   (33) 3.2.1. Sensitivity analysis
LN1 1  B1 L1;HP
xN1 x1;HP Fig. 13 shows total exergy losses in the column as a function of
the reux ratios at the bottom and at the top of the high-pressure
An additional condition is to be added while simulating the column, B1 and C respectively. Fixed values of other parameters are
double adiabatic column. The energy balance at the intermediate as follows. Geometric parameters are chosen in such a way to
heat exchanger should be veried (Eq. (34)). minimize pressure drop in the gas ow.

B2 L1;LP hfg;LP CGN;HP hfg;HP (34)

 High-pressure HP 550 kPa.
Mass ow rates of feeds in the two columns are determined
 Low pressure LP 110 kPa.
regarding the energy balance Eq. (34) and the 5000 tons/day liquid
 High-pressure column height HHP 10 m.
product ow rate needed.
 Low pressure column height HLP 10 m.
 Reux ratio at the bottom of the low pressure column
B2 0.8.
3.2. Exergy analysis of thermally coupled columns
 Feed location in the high-pressure column F1 1 m.
 Feed location in the low pressure column F2 9 m.
Exergy losses in this column conguration are due to mass
 High-pressure column diameter dHP 19 m.
transfer in the two columns, the heat transfer in the boiler and in
 Low pressure column diameter dLP 12 m.
the intermediate heat exchanger, and in the air compressor
providing the high-pressure vapor stream.
when B1 increases, the liquid introduced at the top of the low-
Exergy losses in the compressor are:
pressure column decreases. In order to ensure the 5000 tons/day
  product ow rate, Mfeed,LP increases, causing higher exergy losses in
Exloss;comp Mfeed T scomp;out  scomp;in (35)
the low-pressure column and thus higher exergy losses in the
Besides the geometric parameters, the operating parameters to column (Fig. 13(a)).
be optimized in this column conguration are the followings: On another hand, when the reux ratio at the top of the high-
pressure column increases, the liquid introduced to the top of the
 The reux ratio at the HP boiler, B1. low-pressure column increases too. This causes a decrease in the
 The reux ratio at the top of the HP column, C. liquid feed mass ow rate and thus the decrease of total exergy
 The reux ratio at the bottom of the LP column, B2. losses.
 The low pressure, LP. Fig. 14 shows the inuence of the reux ratio B2 at the bottom of
 The high pressure HP. the low-pressure column on total exergy losses, for reux ratios at
the bottom and at the top of the high-pressure column of 0.4. Total
The above listed parameters are the input data of the algorithm exergy losses in the column increase when B2 increases because the
used to simulate the columns and calculate the total exergy losses feed mass ow rates increase.
 J. Rizk et al. / Energy 37 (2012) 417e429
Fig. 13. Inuence of the reux ratios in the high-pressure column on the total exergy losses, (a) effect of the reux ratio at the bottom, (b) effect of the reux ratio at the top.
Fig. 15 shows the effect of pressure levels in the columns on total
exergy losses. Since a constant temperature difference of 5 K is
assumed at the intermediate heat exchanger between the high-
pressure vapor condensation and the low-pressure liquid evapo-
ration, the pressure level of the high-pressure column is calculated
for each value of the low-pressure to ensure this condition. An
increase in the low-pressure column is followed by an increase in
the high-pressure column causing an increase in the compressor
exergy losses. The compressor contribution in total exergy losses in
the double adiabatic column being the highest (50e60%), total
exergy losses will increase due to the increase in the compressor
exergy losses.
For the double adiabatic column, the geometric parameters
(diameters and heights of columns and feed locations) present the
same effect as for the simple adiabatic column. Total exergy losses
decrease when the column diameters increase because the pres-
sure drop decreases. The feed location, in both columns, has very
little effect on total exergy losses.
3.2.2. Overall optimization of the column
Several possible combinations of geometric and operating
parameters are considered and total exergy losses are calculated for
each combination. A decreasing curve is obtained (Fig. 16) and
optimal congurations are deduced. Minimum exergy losses of
31.4 kW/kg of vapor feed are calculated for a conguration where
the low-pressure column operates at 110 kPa and the high-pressure
column at 750 kPa. The columns are 22-m diameter and 10-m
length each. The reux ratios are as follows: 0.1 at the high-
pressure boiler, 0.6 at the high-pressure condenser and 0.8 at the
low-pressure condenser. The vapor feed in the high-pressure
column is introduced at a 1-m height from the bottom of the
Fig. 14. Effect of the bottom reux ratio in the low-pressure column on exergy losses.
425
column and the liquid feed in the low-pressure column at a 9-m
height from the bottom.
4. Diabatic distillation column
The thermal coupling between adiabatic columns includes an
intermediate heat exchanger where low-pressure liquid evaporates
and high-pressure vapor condensates. Further thermodynamic
improvements could be achieved in the distillation processes by
distributing this thermal coupling over the height of the columns.
Instead of increasing the high-pressure in the column in order to
ensure a 5-K pinch at the intermediate heat exchanger, a lower
value of the pressure could be tested if the heat exchange between
the columns is still achieved through wall contacts of the two
columns.
The rst approach for creating diabatic ows in distillation
columns was by adding sequential heat exchangers all through the
column height [5,7,22,25,26]. Fig. 17 shows heat exchangers, adding
heat to the stripping section of the column, and removing heat from
the rectifying section. In this conguration, the exergy destruction,
located principally in the boiler and the condenser, is distributed on
trays [5]. As a result, the diabatic column destructs half the exergy
a conventional column would destroy for the same separation.
Rivero et al. [22,25] searched for the optimal allocation of the
sequential heat exchangers leading to the best thermodynamic
performances, and calculated 39% exergy loss decrease in the dia-
batic column, compared to the conventional column.
Recently, a new type of diabatic distillation column was pre-
sented. Known as internally heat-integrated distillation column, it
has the same approach of adding and removing heat through the
column height, but the stripping section and the rectifying section
Fig. 15. Effect of pressures in the columns on total exergy losses.
426 J. Rizk et al. / Energy 37 (2012) 417e429

Fig. 16. Total exergy losses in different congurations of double adiabatic columns.

are put into contact and form a heat exchanger, where heat is
transferred from the hot source (the HP rectifying section) to the
cold sink (the LP stripping section) [2,3,27]. The ideal conguration
of the internally heat-integrated columns contains neither boiler
nor condenser. Therefore, it reduces the total exergy losses in the
column and increases its efciency.
In this paper, an ideal internally heat-integrated distillation
column is considered and represented in Fig. 18.
The simulation of the diabatic column operation differs from the
previous studied adiabatic columns, where an industrial packed
column was considered. The conservation equations should be
adapted to take into account the additional heat transfers through
the column.
4.1. Simulation of the diabatic distillation column
In order to simulate the diabatic column, we assume that the
distillation process is realized in a heat exchanger which surface
contact area is sufcient to exchange heat between the low-
pressure mixtures and the high-pressure mixtures.
In this distillation conguration, mass ow rates of the liquid
and vapor mixtures vary more markedly, due to condensation of
Fig. 18. Ideal internally heat integrated distillation column.
the high-pressure vapor and the evaporation of the low-pressure
liquid. On the nite volume i of Fig. 3, an additional variable, the
heat transferred, is to be added.
To calculate the amount of heat transferred, an overall heat
transfer coefcient is rst correlated. It takes into account
a convective heat transfer coefcient during the condensation of
vapor mixtures at high pressure, a convective heat transfer coef-
cient during the evaporation of liquid mixtures at low-pressure and
a conductive heat transfer coefcient through the heat exchanger
wall. Correlations for convective heat transfer coefcients are
detailed in [28]. Thome correlation for evaporation in the low-
pressure column and Ackers correlation for the condensation in
the high-pressure column are used in this study. When the overall
heat transfer coefcient is evaluated, the heat exchanged is calcu-
lated as a function of the temperature difference between the two
columns.
The conservation equations with heat transfer are replacing Eqs.
(19)e(22), and give Eqs. (36)e(39).

xi  x*i Qi
Li1 xi1  Li xi  kL;i Si xeq (36)
Vi hfg;eq

 Qi
Gi1 yi1  Gi yi kG;i Si Pi y*i  yi xeq (37)
hfg;eq

xi  x*i hfg;N2 xi  x*i Qi

Li1  Li  kL;i Si kL;i Si (38)
Vi hfg;O2 Vi hfg;eq

 hfg;N2
* 
Gi1  Gi kG;i Si Pi y*i  yi  kG;i Si Pi yi  yi
hfg;O2
Q
 i (39)
hfg;eq

In Eqs. (36)e(39), hfg,eq is the heat of vaporization of the mixture in

Fig. 17. Heat exchanger sequence in a HIDiC.
equilibrium with the vapor at high pressure or with the liquid at
low-pressure. xeq is the nitrogen molar fraction in the mixture.
The pressure drop in the heat exchanger considered for the
diabatic distillation is calculated by applying the corrections to the
vapor velocity and to the friction factor in the linear pressure drop
equation [13].
 J. Rizk et al. / Energy 37 (2012) 417e429 427

The method of simulation of the diabatic distillation process in
a heat exchanger is as follows:
1. The feed mass ow rate and molar composition being known,
assume a linear variation of the liquid and the vapor mass ow
rates and compositions.
2. Assume a linear variation of the pressures.
3. Calculate liquid saturation temperatures using REFPROP 8.0
with compositions and pressures from 1 and 2.
4. Calculate thermodynamic properties of liquid and vapor
mixtures on both pressure levels.
5. Calculate pressure drops using properties from 4.
6. Calculate corrected thermodynamic properties using pressures
from 5.
7. Calculate heat transfer coefcients and heat exchanged.
8. Calculate mass transfer coefcients.
9. Write 6N equations for the high-pressure column and resolve
the system AX B.
10. Using L1 and x 1 at high pressure, write 6N equations for the
low-pressure columns and resolve the system AX B.
11. Calculate the relative differences for both columns.
12. If maximum relative difference higher than 104, repeat from
step 4. Else, proceed to exergy analysis.
determine the number of passes needed to ensure the 5000 tons
per day of the liquid product.
Before proceeding to the general optimization of the process,
a sensitivity analysis is achieved, to study the effect of each
parameter on total exergy losses in the column, producing 5000
tons per day of oxygen rich liquid at low-pressure.
4.2.1. Sensitivity analysis
The effect of each parameter on total exergy losses in the dia-
batic column is now studied, by choosing xed values for other
parameters.
First, the effect of pressure levels on each side of the column is
studied. Fig. 19 shows the effect of the pressure ratio between the
two sides of the heat exchanger on the total exergy losses.
Fig. 19(a) represents the total exergy losses per kg of high-
pressure vapor feed and Fig. 19(b) represents them per kg of
low-pressure liquid product. The xed values of other parameters
are xed to:
 H 10 m
 dLP 44 mm
 dHP 44 mm
 Mfeed 0.1 kg/s/pass.

When studying the dependence of the total exergy losses as

4.2. Exergy analysis of the diabatic column a function of the pressure ratio for a given high-pressure feed, an
increasing curve is observed. In fact, exergy losses in the
Exergy losses in the diabatic distillation column are due to compressor are proportional to the high-pressure level and those in
irreversibility on each nite volume and to the air compression. expansion valve are proportional to the pressure ratio. They both
On each nite volume, exergy losses are calculated by an cause an increase in total exergy losses for a given feed mass ow
exergy balance, with the heat transfer taken into account rate.
(Eq. (40)). On another hand, when the pressure ratio increases, the amount
of heat exchanged between the high- and the low-pressure uxes
 
Tref increases, and the amount of low-pressure liquid collected at the
Exloss
i L ExL
i1 i1 G ExV
i1 i1  L Ex
i i
L
G Ex
i i
L
Qi 1  bottom of the low-pressure column for a given vapor feed mass
Ti
ow rate increases. As a result, the number of passes needed to
(40)
ensure 5000-ton per day of pure oxygen decreases and total exergy
Optimal geometric and operating parameters, leading to losses too (Fig. 19(b)).
minimum exergy losses, are calculated. Parameters to be optimized Graphs of Fig. 20 show the effect of geometric parameters
are: (height of the column and diameter of the low-pressure pass).
When the column height increases, the pressure drop in the
 Column height, H. columns and the quantity of heat exchanged increase. Both
 Hydraulic diameter at low-pressure, dLP. pressure drop and heat increase exergy losses (Fig. 20).
 Pressure ratio HP/LP. An increase in the column diameter leads to a decrease in the
pressure drop and to improvement in mass transfers between the
In order to simplify the optimization procedure, the column is liquid and vapor mixtures (higher interfacial surfaces). This
considered as a two-pass heat exchanger, one pass at each pressure explains the decrease in exergy losses when the diameter increases,
level. For each mass ow rate of the feed, simulation results Fig. 20(b).

Fig. 19. Exergy losses as a function of the pressure ratio, (a) per kg of vapor feed, (b) per kg of liquid product.
428 J. Rizk et al. / Energy 37 (2012) 417e429

Fig. 20. Exergy losses as a function of geometric parameters, (a) as a function of the column height, (b) as a function of low-pressure pass diameter.

4.2.2. Overall optimization of the column 6. Concluding remarks

As for adiabatic columns, exergy losses are calculated only when
the desired purity of the liquid product is achieved. A decreasing
curve is obtained (Fig. 21), representing the exergy losses per kg of
feed mass ow rate. A 20.7-m diameter, 10-m height column,
operating at a high pressure of 450 kPa and a low-pressure of
150 kPa, leads to the minimum exergy losses of 24 kW/kg of feed.
By comparing the optimal values in Fig. 21 and those plotted in
Fig. 16, a relative decrease of 23% in the total exergy losses are
calculated between the optimal conguration of the double adia-
batic column and the optimal conguration in the diabatic column.
In this study, optimal congurations for minimal total exergy
losses of three types of cryogenic distillation columns are calcu-
lated. The columns considered for the comparison are the simple
adiabatic column, the double adiabatic column, and the internally
heat-integrated column (diabatic column). The double diabatic
column presents 23% less exergy losses than the double adiabatic
column. It should be noted that a general optimization of the global
process including the distillation columns will lead to more specic
results and comparisons.

5. Comparison of the three types of cryogenic distillation Notation

columns

Performances of double adiabatic columns and diabatic columns aP Packing surface area (m2/m3)
are compared regarding total exergy losses calculated in optimal bP Corrugation base length (m)
congurations. A total of 23% decrease in total exergy losses are B Reux at the boiler
calculated when replacing the local intermediate heat exchanger of C Reux at the condenser
the double adiabatic column by an overall heat transfer between d Column diameter (m)
the columns. This decrease is due mainly to the decrease in the dh Hydraulic diameter (m)
pressure in the high-pressure column (from 750 kPa to 450 kPa), D Diffusion coefcient (m2/s)
decreasing exergy losses in the compressor. It is also due to the Ex Exergy (W)
decrease in irreversibilities in the column due to heat transfers. f Friction factor
The comparison with the simple adiabatic column cannot be F Feed location (m)
made on a similar basis since the products of the simple adiabatic h Enthalpy (J/kg)
column distillation are different from those of the double column hfg Heat of vaporization (J/kg)
(presence of a high-pressure nitrogen rich vapor product in the hL Liquid holdup
double columns). Besides, the simple adiabatic column has no hP Corrugation height (m)
compressors and the production of the cold source for the hpe Height of the packing element (m)
condensation at the top is not considered. H Column height (m)
k Mass transfer coefcient
K Henry constant
lG,pe Length of the vapor ow channel within a packing
element (m)
L Liquid mass ow rate (kg/s)
m_ Mass ow rate (kg/s)
P Pressure (bar)
Q Heat (W)
s Entropy (K/kg.K)
sP Corrugation side length (m)
Si Interfacial surface (m2)
T Temperature (K)
v Velocity (m/s)
V Vapor mass ow rate (kg/s)
Vm Molar volume (m3/mol)
x Nitrogen molar fraction in the liquid
Fig. 21. Exergy losses in the diabatic column for different congurations. x0 Oxygen molar fraction in the liquid
 J. Rizk et al. / Energy 37 (2012) 417e429 429

x* Nitrogen molar fraction in the liquid at the interface [3] Ho TS, Huang CT, Lin JM, Lee LS. Dynamic simulation for internally heat-
integrated distillations columns (HIDiC) for propylene-propane system.
y Nitrogen molar fraction in the vapor
Computers and Chemical Engineering 2009;33:1187e201.
y0 Oxygen molar fraction in the vapor [4] Yong PS, Moon HM, Yi SC. Exergy analysis of cryogenic air separation process
y* Nitrogen molar fraction in the vapor at the interface for generating nitrogen. Journal of Industrial and Engineering Chemistry 2002;
W Work (W) 8(6):499e505.
[5] Le Goff P, Cachot T, Rivero R. Exergy analysis of distillation processes.
Chemical Engineering Technology 1996;19:478e86.
[6] Agrawal R, Fidkowski ZT. Are thermally coupled distillation columns always
Greek letters thermodynamically more efcient for ternary distillations. Industrial and
a Corrugation angle with the horizontal ( ) Engineering Chemical Research 1998;37:3444e54.
g Contact angle ( ) [7] De Koeiger G, Rivero R. Entropy production and exergy loss in experimental
distillation columns. Chemical Engineering Science 2003;58:1587e97.
d Liquid lm thickness (m)
[8] Olujic Z, Kamerbeek AB, De Graauw J. A corrugation geometry based model for
Porosity efciency of structured distillation packing. Chemical Engineering and Pro-
h Efciency cessing 1999;38:683e95.
r Density (kg/m3) [9] Copigneaux P. Distillation. Absorption e Colonnes garnies, Techniques de
lIngnieur, J2626.
m Viscosity (Pa s) [10] Rocha JA, Bravo JL, Fair JR. Distillation columns containing structured pack-
4 Fraction of the triangular ow channel occupied by liquid ings: a comprehensive model for their performance. 2. Mass transfer model.
xGL Liquid/vapor friction factor Industrial and Engineering Chemical Research 1996;35:1660e7.
[11] Olujic Z. Compute friction factors fast for ow in pipes. Chemical Engineering
1981;88(25):91e3.
Non-dimensional numbers [12] Ratheesh S, Kannan A. Holdup and pressure drop studies in structured
packings with catalysts. Chemical Engineering Journal 2004;104:45e54.
Fr Froude number [13] Woerlee GF, Berends J, Olujic Z, de Graauw J. A comprehensive model for the
Re Reynolds number pressure drop in vertical pipes and packed columns. Chemical Engineering
Sc Schmidt number Journal 2001;84:367e79.
[14] Fernandes J, Lisboa PF, Simes PC, Mota JPB, Saatdjian E. Application of CFD in
Sh Sherwood number
the study of supercritical uid extraction with structured packing: wet
We Weber number pressure drop calculations. The Journal of Supercritical Fluids 2009;50:61e8.
[15] Lallemand A. Thermodynamique applique, premier principe, nergie,
enthalpie. Techniques de lIngnieur BE 8005-1.
Super- and subscripts [16] Hernandez-Gaona CG, Cardenas JC, Segovia-Hernandez JG, Hernandez S, Rico-
B Boiler Ramirez V. Second law analysis of conventional and nonconventional distil-
boil Boiler lation sequences. Chemical and Biochemical Engineering Quarterly 2005;19:
235e41.
C Condenser [17] Paiva CR, De Lacarda AI. Chemical process improvements by exergy analysis.
comp Compressor Fourth Mercosur Congress on Process Systems Engineering; 2005.
cond Condenser [18] Petlyuk FB, Platonov VM, Slavinskii DM. Thermodynamically optimal method
for separating multi-component mixtures. International Chemical Engi-
e Effective neering 1965;5(3):555e61.
eq Equilibrium [19] Jimenez A, Ramirez N, Castro A, Hernandez S. Design and energy performance
G Vapor of alternative schemes to the Petlyuk distillation system. Institution of
Chemical Engineering IChemE 2003;81:518e24.
HP High pressure
[20] Hernandez S, Jimenez J. Design of energy-efcient Petlyuk systems.
i Relative to the nite volume (i) Computers and Chemical Engineering 1999;23:1005e10.
L Liquid [21] Hernandez S, Jimenez J. Design of optimal thermally-coupled distillation
systems using a dynamic model. IChemE; 1996.
lam Laminar
[22] Rivero R. Exergy simulation and optimisation of adiabatic and adiabatic binary
LP Low pressure distillation. Energy 2001;26:561e93.
ref Reference [23] Chen Z, Henson MA, Belanger P, Megan L. IEEE Transactions on Control
rv Relative velocity Systems Technology; 2009.
[24] Ha B, Ramon S. Cryogenic distillation system for air separation. United States
s Supercial Patent; 2001.
turb Turbulent [25] Jimenez ES, Salamon P, Rivero R, Rendon CC, Hoffmann KH. Optimal allocation
of heat exchanger inventory in a serial type diabatic distillation column.
Proceedings of ECOS; 2004:179e87.
References [26] Dernane C, Laouir A. Etude comparative entre une distillation conventionnelle
et une distillation diabatique. Revue des Energies Renouvelable CISM; 2008. p.
[1] Cornelissen RL, Hiesg G. Exergy analysis of cryogenic air separation. Energy 123e34.
Conversation Management 1998;39:1821e6. [27] Jana AK. Heat integrated distillation operation. Applied Energy 2010;87:
[2] Gadalla MA. Internal heat integrated distillation columns (iHIDiCs) e new 1477e94.
systematic design methodology. Chemical Engineering Research and Design [28] Hamdar M. Design and optimization of an efcient indirect air-conditioning
2009;87:1658e66. system for train carriages. Ph.D. Mines-ParisTech; 2008.