Energy 35 (2010) 15051514

Contents lists available at ScienceDirect

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

Design of internally heat-integrated distillation column (HIDiC):

Uniform heat transfer area versus uniform heat distribution
B. Suphanit*
Department of Chemical Engineering, Faculty of Engineering, King Mongkuts University of Technology Thonburi, Pracha Utit Rd., Tungkru, Bangkok 10140, Thailand

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

Article history: The internally heat-integrated distillation column (HIDiC) is a complex column conguration which is
Received 29 April 2009 more energy efcient than the equivalent conventional column or the distillation column with direct
Received in revised form vapor recompression scheme (VRC). Exploiting the heat integration between two diabatic sections
9 November 2009
operating at different pressures of the HIDiC can greatly enhance the energy performance of the system.
Accepted 5 December 2009
On the other hand, the design and optimization of HIDiC is more difcult than those of the conventional
Available online 28 December 2009
distillation column or the column with VRC. The former involves many design parameters, and the most
critical one is the pressure ratio between both diabatic sections. However, the heat distribution along the
Keywords:
Heat-integrated distillation column diabatic sections is also another signicant factor not yet thoroughly investigated. In this work, two
HIDiC typical distribution schemes, i.e. uniform heat transfer area and uniform heat distribution, are studied by
Uniform heat transfer area applying a novel approach to solve the simulation problem in Aspen Plus 2004.1. The comparison of both
Uniform heat distribution distributing schemes is discussed via two widely-used case studies, namely benzene-toluene separation
and propylene-propane splitter.
2009 Elsevier Ltd. All rights reserved.

1. Introduction energy requirement along both sections can then be saved. As

shown in Fig. 1a, the rectifying and the stripping sections are
As distillation is one of the most energy-intensive units in separated by the feed tray. Instead of having heat rejection only
chemical process industries, many efforts have been focused on the through the main condenser and heat supply through the main
development of distillation systems and distillation equipment in reboiler, the total heat rejection is distributed along the rectifying
order to improve its energy efciency. Many complex distillation section, while the stripping section absorbs this amount of heat
congurations have been studied and proposed to improve the with similar distribution along the column. A compressor and
energy efciency. For example, the dividing-wall distillation a throttling valve are used to manipulate the pressure difference
column or the Petlyuk column is a complex distillation scheme between both column sections. The rectifying pressure must be
which has been successfully commercialized [1] after several sufciently high to provide a positive temperature driving force
decades of research [2]. between both sections for feasible heat transfer along the columns.
More recently, another distillation scheme called the internally This internal heat integration between column sections can be
heat-integrated distillation column (HIDiC) has received extensive achieved using a heat transfer equipment at any stage location. The
attention. Unlike the Petlyuk column, the key idea of the HIDiC design without any thermal utility (as in Fig. 1a) is called the ideal
leading to considerable energy savings is the combination of direct HIDiC. This scheme requires only electricity or mechanical power
vapor recompression scheme (VRC) and the heat integration for compressor. However, a trim-condenser and/or trim-reboiler
between two diabatic sections. The diabatic section is a column may be installed at the top of the rectifying and/or the bottom of
section which allows heat transfer to enter or leave its stages. Its the stripping sections, respectively, to enhance the process
purpose is to distribute exergy loss from the main condenser or operational exibility [3]. The general HIDiC scheme with any trim-
reboiler to all stages in the section, hence reducing the main utility condenser and/or trim-reboiler is known as the internally heat-
load and the overall exergy loss of the section. If two diabatic integrated distillation column (Fig. 1b).
sections with opposite directions of heat transfer are integrated, the The advantage of the HIDiC over the conventional column in
terms of energy efciency was shown by Mah et al. [4]. The system
was then known as the distillation with secondary reux and
* Tel.: 66 2 470 9222x401; fax: 66 2 428 3534. vaporization (SRV). Moreover, the exergy loss analysis studies of the
E-mail address: [email protected] HIDiC were carried out which showed that the HIDiC had high

0360-5442/$ see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2009.12.008
1506 B. Suphanit / Energy 35 (2010) 15051514

Compressor Compressor
Distillate

Feed Feed
Distillate

Valve Rectifying Valve Rectifying

Section Section

Stripping Stripping
Section Section

Bottom
Bottom

a b
Fig. 1. The ideal HIDiC (a) and the internally heat-integrated distillation column (b).

energy efciency, judging from its low exergy loss value [5,6]. The carried out with either constant values of heat transfer area [47] or
HIDiC was also shown to give a substantial energy savings of constant values of heat transfer rate [8] along the column section.
around 3050% in the separation of various mixtures when The simulation carried out by Gadalla et al. [15,16] showed that
compared to the conventional column [7]. In comparison with VRC, designing the HIDiC with some non-integrated stages or with non-
the HIDiC scheme can achieve up to 20% in the reduction of the total uniform heat transfer scheme along the column section could yield
annualized cost due to its lower compression ratio, leading to better performance.
a smaller-sized compressor [8,9]. The HIDiC simulation with either xed heat transfer area or
The operability and control of the HIDiC have also been studied xed heat transfer rate may lead to convergence problems or may
extensively in order to bring this scheme into industrial applica- result in the design with partial heat integration between both
tions. Different control algorithms were explored to identify the sections. Here, the HIDiC with partial heat integration is designated
most suitable scheme for the ideal HIDiC [10]. In addition, the ideal as the HIDiC which cannot achieve the maximum possible heat
HIDiC scheme with heat integration between the distillate and the recovery between column sections, i.e. the resulting HIDiC will
feed was also studied [11,12]. The application of HIDiC is best suited need both condensing and reboiling loads. On the other hand, the
for the separation of close-boiling mixtures and low operating HIDiC exhibiting maximum possible heat recovery between
temperatures, such as trans-2-butene/cis-2-butene, propylene/ column sections will be called the full HIDiC. To achieve full heat
propane, or ethylene/ethane separations [4]. As a result, most integration between both sections, either the heat transfer area or
research on the control aspects was focused on these kinds of case the heat load along the column section could not be xed during
studies, e.g. benzene-toluene separation and propylene-propane the simulation but must instead be allowed to vary.
splitter [1013]. These studies have led to a better understanding of In this study, a new approach to the simulation of the full HIDiC
the dynamic behavior and the controllability of the HIDiC. is presented using the commercial simulator, Aspen Plus 2004.1.
However, bench-scale and pilot-scale testing must be carried out Two typical patterns of heat distribution in the HIDiC design will be
before commercialization can begin. From a bench-scale experi- studied and compared. The rst is to transfer heat between column
mental test [14], it was shown that the operation of the ideal HIDiC sections according to the temperature driving force along both
was feasible and the energy consumption was about 40% less than sections. The location with a high temperature driving force will be
that of the conventional column in the separation of benzene and allocated with a large amount of heat transfer when compared to
toluene. Many pilot or even industrial-scale projects of HIDiCs are others with lower driving forces. This heat distribution scheme
currently underway in Japan and in the Netherlands [1]. Different results in uniform heat transfer area along the column section. The
types of the HIDiC column congurations are now being developed. other case is to consider uniform heat distribution along the
For example, the vertical shell-and-tube column with rectifying column section. In this case, the area requirement at each stage
section as multiple tubes placed in parallel inside the stripping shell location will be different and depends on the temperature driving
section was successfully demonstrated in Japan, while the force at that particular location.
concentric tray column with heat panel is now focused in the
Netherlands. The application of plate-n type devices in the HIDiC 2. Design and simulation approach
is also another approach being explored. The study has now
entered the industrial-scale testing phase in Japan [1]. In the design and optimization of any process, the total annu-
Although research in the HIDiC has reached the stage of alized cost (TAC) is generally used as the indicator to determine the
industrial-scale testing, the number of design methodologies optimum design condition. For HIDiC, there are many design vari-
developed so far was quite small and not clearly established. Some ables involved in the design and optimization. The major contrib-
of these studies were based on user-written models [47], while uting factor affecting the TAC in the design of HIDiC is the pressure
others used the commercial process simulators such as Aspen Plus difference or the pressure ratio between column sections [3]. As the
or Aspen HYSYS [8,15,16]. The HIDiC design and simulation were electricity for driving the compressor is several times more
 B. Suphanit / Energy 35 (2010) 15051514 1507

expensive than the thermal utility, the value of pressure ratio heat distribution, are investigated in more detail using a new
should be kept as low as possible to minimize the electricity cost. approach to the simulation problem. The simulation procedure is
However, the vapor owing through the compressor is another based on the modeling in Aspen Plus 2004.1. Each column section is
signicant factor which can affect the power requirement in the represented by a RADFRAC model. Other auxiliary pieces of
compressor. In the HIDiC, the vapor ow in each section varies equipment, such as compressor or throttling valve, are also
signicantly along the column section. The vapor ow decreases modeled using standard Aspen Plus blocks.
from the bottom to the top of the rectifying section and increases as The procedure is briey summarized in Fig. 2. For any given
it ows up towards the top of the stripping section. Generally for feed condition, the number of stages, the location of stages being
HIDiC, the vapor ow at the top of the stripping section is larger matched up, the pressure in each section, and the top & bottom
than the average vapor ow in a typical VRC scheme [8]. Although product specications must be predetermined. The heat distrib-
the low pressure ratio can provide benets for the HIDiC, the large uting pattern along the column section is also selected. The
vapor ow through the compressor has a negative impact on the desired simulation results will be the values of heat loads at any
compressor power requirement. The amount of vapor ow through stage. However, these values of heat loads are the required input
the compressor, i.e. the vapor ow at the top of the stripping variables in Aspen Plus; therefore some iterations are necessary,
section or the bottom of the rectifying section, depends on the total provided that good estimates are initially specied. To obtain the
amount of heat transfer between both column sections. A large initial estimates of heat loads at every stage, the procedure starts
amount of total heat transfer not only requires a large heat transfer with the simulation of HIDiC conguration without heat integra-
area, but it also results in a high value of vapor ow entering the tion between column sections. The total amount of heat rejected
compressor. Therefore, the amount of heat transfer between both from the rectifying section (QR) will be specied as the calculated
column sections, though having no effect on the thermal utility condensing load, and the total amount of heat required in the
requirement, should be kept as low as possible. This could be stripping section (QS) as the reboiling load. Either one of these
attained by good heat distribution along the column sections. The values will be set as the initial value of the total amount of heat
appropriate heat distribution along the column section is very transfer between column sections (QT). If the value of QR is greater
crucial in obtaining the diabatic section with high thermodynamic than QS, the maximum heat integration between sections will be
efciency. With proper heat distribution, the column exergy loss limited by the total amount of heat required in the stripping
will be low, thereby resulting in low column vapor ows. Failing to section. Hence, the HIDiC in this case requires no hot utility or
do so will result in a large energy penalty on the total amount of external reboiler. The initial value of QT is then set as QS. On the
heat transfer along the column section, which subsequently leads other hand, the initial value of QT will be set as QR if the value of
to a high vapor ow through the compressor. QR is less than QS. In other words, the amount of heat rejected
The suitable scheme of heat distribution in HIDiC is case- from the rectifying section (QR) determines the maximum heat
specic. Allowing the heat transfer at any particular stage to vary integration between both sections. Having set up the initial value
independently without any distributing pattern during the design of QT, the heat distribution along the column sections can be
and optimization could be a daunting task. Here, two widely-used estimated depending on the heat distributing scheme under
distributing patterns; i.e. uniform heat transfer area and uniform consideration.

Fig. 2. Simulation procedure for HIDiC with either uniform heat transfer area or uniform heat distribution scheme.
1508 B. Suphanit / Energy 35 (2010) 15051514

For the HIDiC with uniform heat transfer area, the heat load (Qi) the pressure ratio between column sections or the rectifying
will be distributed according to the temperature driving force at pressure when xing the stripping pressure. This simple optimi-
any stage location (DTi) along the column. Basically, the overall heat zation can be carried out by a Sensitivity block in Aspen Plus. In the
transfer coefcient (U) is approximately constant along the column preliminary design level, the objective function is the total annu-
sections. The heat load at any stage location (Qi) can then be esti- alized cost which basically consists of the annualized equipment
mated using Eq. (1). cost and the annualized utility cost. To perform the analysis at this
0 1 level, the equipment cost, utility cost and the economic condition
must be specied in order to evaluate the TAC value. The installed
B Q C equipment cost estimates are calculated as given by Douglas [17].
B C
Qi DTi B n T C (1)
@P A These estimates are escalated to the present condition using M&S
DTi Index. In this study, the value of 1463.2 averaged for process
i1
industries in the 1st quarter of 2008 is applied [18].
In case of large variation in DTi along the column section, the Column sizing and costing basis by Olujic et al. [8] is also applied
overall heat transfer coefcient may be, however, not constant, but here. The material of construction for both shell and tray is the
more accurately depends on the value of local temperature differ- carbon steel. The column sizing is estimated using the tray sizing
ence. If the relationship of Ui as a function of DTi can be correlated option in RADFRAC block. It was assumed that all trays in both
from experimental test, Eq. (2) can be used instead to estimate the sections are sieve tray with 1.5-ft tray spacing. The column diam-
heat distribution along the column. Similar to Eq. (1), the uniform eter sizing is designed at 80% ooding using Fair correlation. During
heat transfer area can still be obtained. this conceptual study, all stages are theoretical trays, and the
0 1 column height is calculated accordingly. The column is constructed
as a concentric column. The column cost is then estimated from
B Q C
B C two-column shells. The diameter of the inner column, i.e. the
Qi Ui DTi B n T C (2)
@P A rectifying section, is not constant as a result of large variation in the
Ui DTi
i1 vapor ow along the section. Hence, the diameter used in this
column sizing is the average value of all diameters calculated in this
For the HIDiC with uniform heat distribution, the heat load at section. For the outer column, after accounting for the rectifying
any stage (Qi) can be readily estimated by dividing QT with the and the stripping tray area required at any stage location, the outer
number of heat transfer stages (n), as shown in Eq. (3). diameter can be determined. The maximum diameter value along
  the section is chosen in the cost estimation of the outer shell. Due
QT
Qi (3) to the complexity of the concentric column, a factor of 1.5 is applied
n
to the overall shell cost. Moreover, the tray costing in the stripping
After initializing the heat distribution along the column section, section is taken as a full cross-section tray, not an annular tray as
the simulation of HIDiC can then be carried out. The heat streams in built [8].
Aspen Plus are used to represent the heat transfer between the The calculation of the heat exchanger area depends on each type
corresponding stages from both sections. However, the bottom of heat exchanger. The heat transfer between stages in both
stage duty in the stripping section and the top stage duty in the sections is carried out by the heat panel. The other exchangers, e.g.
rectifying section will be determined by the given product speci- condenser or reboiler, will be typical shell-and-tube exchangers.
cations. Once simulation results have been obtained, the total The value of the overall heat transfer coefcient in the heat panel is
amount of heat integration (QT) will be different from the initial assumed constant at 1000 W/m2 K [8]. The U values and the cost
estimate. The new value of QT is therefore updated under the same coefcient for other types of exchangers can be found in Olujic et al.
consideration earlier. The new heat distribution (Qi) can then be [8]. The compressor is assumed to operate at a 72% isentropic
recalculated with the updated value of QT and DTi from the previous efciency (the default value in Aspen Plus). The available utilities
heat-integrated simulation using Eq. (1), (2) or (3) in a similar way are electricity, LP steam, and cooling water with unit costs of $0.1/
as in the initial estimation step. This update is done by formulating kW.h, $13/t, and $0.03/t, respectively [8]. The evaluation of TAC is
user-dened routines in an Excel Calculator block. All values of Qi on the basis of 8000 operating hours per year, a 7% interest rate, and
will be specied as tear variables and adjusted simultaneously a 10-year equipment lifetime.
using the Broyden method. The iterations proceed until the The installation of the heat panel in the concentric column may
convergence of all values of Qi is achieved. The relative tolerance for be subjected to column physical limitations. The required heat
these tear variables is 0.0001 which is the default value in Aspen transfer area for any stage match could be greater than the
Plus. Once arriving at a converged solution, the heat transfer area at maximum heat transfer area physically allowed in the column. In
any particular stage and the economic performance can be evalu- this study, the maximum possible installed area of the heat panel at
ated, also done using an Excel Calculator block. This approach to the any location either in the rectifying or the stripping section will also
HIDiC simulation is very robust and requires no manual adjustment be calculated. The simulation results are nally checked for any
of the heat distribution values. The HIDiC design with maximum physical area violation along the column. Basically, around the top
possible heat integration can then be obtained at any given pres- region of the concentric column (No. 1 as shown in Fig. 3), the
sure ratio. maximum allowable heat transfer area appears in the annular
section (the stripping section). On the other end of the column, the
rectifying section generally provides larger physical space for heat
3. Case studies panel installation than the stripping section (as region No. 2 in
Fig. 3). The calculation of heat panel area is based on the concept by
In this work, two widely-used case studies, i.e. benzene-toluene Gadalla et al. [16]. The dimension of a heat panel unit is 30 mm in
separation, and propylene-propane splitter, were considered with thickness, 30 cm in height, with the length depending on the radial
both heat distribution schemes. Since the heat distributing pattern space in the column. The clearance between the heat panel tip and
is predetermined before the simulation, the only major parameter the column wall is 30 mm. The downcomer area is specied at 10%
left in the optimization at any given number of stages is therefore of the total cross-sectional area of any particular section. Although
 B. Suphanit / Energy 35 (2010) 15051514 1509

600,000
2
580,000

560,000

540,000
3
520,000

TAC ($/yr)
500,000

480,000 4
460,000

440,000 5 All
6 7
420,000

400,000
1.6 1.8 2 2.2 2.4 2.6 2.8 3
Prect (bar)

Fig. 4. TAC prole over a wide range of rectifying pressures in the HIDiC with uniform
Fig. 3. Possible heat panel installation at different location along the concentric heat transfer area for benzene-toluene separation; the number in circle represents the
column. number of heat transfer stages on each end of column sections.

the installation of the heat panel in the rectifying section has not heat panels at these stages with large heat transfer area are
yet been veried in practice [8], the possible area estimation in this removed. Reducing Prect further, the TAC value starts to increase
section is conceptually determined on the same basis of radial again as a result of decreasing the temperature driving force
structure as applied in the annular space with the assumption that between both sections. In this case, these non-integrated stages
the center downcomer is used. occur around the middle of the column sections, leaving the stages
around the top and the bottom regions of the column section to be
3.1. HIDiC for benzene-toluene separation heat-integrated. This heat-integrated scheme is called the HIDiC
with partially-integrated stages (PIS). The lower the rectifying
The feed mixture in this case study consists of benzene and pressure, the fewer the number of heat transfer stages on each end
toluene with equal mole fractions. The feed condition is at 94  C, of both sections remains. Considering all possible HIDiC with PIS
1 atm with a ow rate of 100 kmol/h. The physical property model over a broad range of rectifying pressure, the optimum HIDiC in this
used in this case is the RK-Soave equation-of-state. The number of scheme should be to operate at Prect of 1.97 bar with 5 heat transfer
stages is estimated from the shortcut model for a conventional stages on each end of both sections. The TAC prole in Fig. 4 also
column at 1.1 R/Rmin. This gives 32 stages (including condenser and shows that each HIDiC with PIS has its Prect value with minimum
reboiler) in the column. Each section in the HIDiC then consists of TAC. However, this value is not actually the optimum condition of
16 possible heat transfer stages. The other cases with different each HIDiC scheme. At this condition, the temperature driving force
number of stages can also be explored; however, to demonstrate around the middle of the column sections is rather tight, hence
the approach, only the HIDiC with 16 stages in each section will be requiring large heat transfer area and resulting in a high heat
considered. The stripping pressure is xed at 1 atm, while the exchanger cost. This driving force can be relaxed by allowing Prect to
rectifying pressure is the design variable. In the column simulation, increase while maintaining the same number of heat transfer
the pressure drop in each section is assumed to be zero. The stages on each column end. Although the cost related to the
product specications are the purities of benzene in the distillate compressor could go up from this action, the net effect after
and toluene in the bottom product, both at 99.5%mol. After setting accounting for the reduction in heat transfer area could be bene-
up the HIDiC model in Aspen Plus, the outlined procedure cial in terms of TAC reduction. The sensitivity analysis of Prect on
described in the previous section can now be followed. In this case TAC is then applied individually to each different HIDiC with PIS,
study, the value of QR is greater than QS, therefore no hot utility and the results of which are shown in Fig. 5. The optimum Prect for any
external reboiler is required in this HIDiC. The only design
480,000
parameter in the economic trade-offs is the rectifying pressure
since the heat distributing pattern and the number of stages are 460,000
already specied. The optimum rectifying pressure which gives
minimum TAC can be easily determined by the sensitivity analysis. 440,000
First, the sensitivity analysis results for the uniform heat transfer
TAC ($/yr)

420,000 All
area scheme are shown in Fig. 4. The TAC prole becomes discrete 6
7
when the rectifying pressure is below 2.08 bar. This pressure is 5
400,000 4
called the minimum rectifying pressure for the HIDiC with fully- 3
2
integrated stages (Prect,min). In the region with pressure above this 380,000 1
value, the heat transfer at all stages are possible due to a positive
360,000
temperature driving force between both sections, and the optimum
rectifying pressure is found to be 2.36 bar. However, in the pressure 340,000
region below Prect,min, some corresponding stages between both 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5
sections may exhibit negative temperature differences. These Prect (bar)
stages must be left without any feasible heat transfer. The number
Fig. 5. Effect of rectifying pressure on the TAC of HIDiC with different number of heat
of heat transfer stages is then reduced, causing an abrupt change in transfer stages and uniform heat transfer area for benzene-toluene separation; the
the TAC prole. The TAC value begins to drop sharply as some number in circle represents the number of heat transfer stages on each end of column
corresponding stages become non-integrated stages because the sections.
1510 B. Suphanit / Energy 35 (2010) 15051514

560,000 Table 1
Optimum results under column physical constraint of various HIDiCs with uniform
540,000
heat transfer area for benzene-toluene separation.
520,000
No. of heat transfer stages on each end. Prect (bar) TAC ($/yr)
500,000
1 12.45 1,611,841
480,000
2 7.74 982,560
TAC ($/yr)

460,000 3 4.85 657,405

440,000 4 3.82 540,422
3 All 5 3.5 500,980
420,000 7
2 6 6 3.19 466,421
400,000 1 7 3.01 449,232
5
380,000 4 8 (All) 2.94 445,729
360,000
340,000
1.7 2.2 2.7 3.2 3.7 When taking the column physical constraint into consideration,
Prect (bar) the optimum HIDiC schemes with uniform heat transfer area under
the column physical constraint in this case study are shown in Table
Fig. 6. Effect of rectifying pressure on the TAC of HIDiC with different number of heat
transfer stages and uniform heat distribution for benzene-toluene separation; the
1. The optimum HIDiC under the physical constraint with uniform
number in circle represents the number of heat transfer stages on each end of column heat transfer area and one heat transfer stage on each end of the
sections. section would be at Prect of 12.45 bar. In spite of satisfying the
physical constraint, this design condition should not be selected
since it is very far away from the previously determined optimum
HIDiC with PIS scheme can then be obtained. Since the major cost point, and the TAC is also much higher. From Table 1, the optimum
contribution of this system is the cost related to compression, the HIDiC under the physical constraint with uniform heat transfer area
scheme illustrating close integration between both sections, i.e. is the HIDiC with fully-integrated stages operating at Prect of
only one heat transfer stage on each end of both sections, is 2.94 bar. The Prect and TAC of this HIDiC are however higher than
therefore the optimum scheme. those of the optimum HIDiC without physical constraint. Therefore,
Similarly, for the uniform heat distribution scheme, the sensi- the optimum HIDiC with uniform heat transfer area and one heat
tivity analysis of Prect on TAC can be applied individually to any transfer stage on each end is still of much interest, even though it
HIDiCs with different number of heat transfer stages. The results becomes impractical due to the physical limit in the concentric
are shown as in Fig. 6. Unlike the uniform heat transfer area column. Hence, rather than implementing the concentric column
scheme, the TAC proles of some HIDiCs with PIS do not exhibit the for the optimum HIDiC with one heat transfer stage on each end,
convex shapes. The congurations are not feasible in some pressure two separated columns with external heat exchangers at the top
regions. Again, the optimum conguration is the HIDiC with only and bottom ends of each column (Fig. 8) can be chosen as
one heat transfer stage on each end of both sections. The major cost a substitute. The TAC of this conguration is $342,320/yr which is
contribution of the system is also the compressor-related cost as in very close to the TAC of the equivalent concentric column at Prect of
the case of the HIDiC with uniform heat transfer area. The optimum 1.88 bar.
conditions of all HIDiC schemes are concluded in Fig. 7. For every
HIDiC with PIS case, the scheme with uniform heat transfer area
3.2. HIDiC for propylene-propane splitter
can perform better in terms of TAC than the equivalent one with
uniform heat distribution. The overall optimum is therefore the
The application of HIDiC scheme with propylene-propane
HIDiC with uniform heat transfer area scheme and one heat
splitter in a typical olen process has shown promise when
transfer stage on each end of the section. The optimum operating
Prect is at 1.88 bar. At this optimum condition, the TAC is minimum
at $345,867/yr. The major cost contribution is the electricity cost
which accounts for 38% of TAC. The compressor power requirement Compressor
is around 164 kW. The required heat transfer areas for the top and
Feed
bottom stage locations are equal at 197.24 m2 each.
Stage 1
Distillate
Uniform heat transfer area Uniform heat distribution
500,000

400,000

Rectifying
T AC ($ /yr)

Valve
300,000
Section

200,000
Stripping
Section

100,000

0
Stage 16
1 2 3 4 5 6 7 8 Bottom
No. of heat transfer stages on each end

Fig. 7. Comparison of the optimum TAC of HIDiC with different number of heat
transfer stages on each end of the sections and different distributing schemes; the Fig. 8. The optimum HIDiC structure with only one heat transfer stage on each end of
number on the bar represents the optimum rectifying pressure in each case. the sections for benzene-toluene separation.
 B. Suphanit / Energy 35 (2010) 15051514 1511

Distillate
Feed Stage 1
Stage 1 Distillate

1
1

Bottom Feed

Stage 62 Stage 110

Valve 2

Compressor
2 Valve 3

Compressor Bottom
Stage 171

Stage 171
Fig. 11. The concentric HIDiC with bottom-integrated conguration for PP splitter; the
number in circle represents the column section number for sizing and costing.
Fig. 9. The concentric HIDiC with top-integrated conguration for PP splitter; the
number in circle represents the column section number for sizing and costing.
section, it is possible to integrate the stripping section with the top
location of the rectifying section or somewhere further down until
compared to the VRC scheme already in use [8]. In this study, the the bottom of the rectifying section is reached. It was shown that
effect of heat distributing scheme on the TAC is further investi- the top-integrated scheme shows the best performance when
gated, as previous works did not systematically explore this issue. compared to the bottom-integrated scheme [15]. The comparison
The feed composition and the condition are the same as used by was however carried out with constant heat distribution. Some
Olujic et al. [8]. The feed mixture consists of 52%mol propylene and non-uniform heat distribution schemes were also shown to give
47%mol propane, while the remaining is mostly i-butane. This feed better cost performance [16]. Nevertheless, the optimum condition
enters the system at 11.2 bar which is the same as the pressure at was not thoroughly investigated. Here, the sensitivity of rectifying
the top of the stripping section. The feed ow rate is at 112,000 kg/h pressure on TAC will be analyzed on three integrated congura-
with 37%vapor. The Peng-Robinson property model is used to tions, namely top-integrated, middle-integrated, and bottom-
predict the uid properties and the vaporliquid equilibria. The integrated, as shown in Figs. 911, respectively, all with both types
pressure drop per stage is xed at 6.2 mbar. The purities of of heat distributing schemes considered in this work. The column
propylene in the products are specied at 99.6%mol in the distillate, cost estimation is different in each integrated conguration. There
and 1.1%mol in the bottom product. The number of stages in the are different numbers of column sections to be evaluated as illus-
rectifying and the stripping sections are 171 and 62, respectively. trated in those gures.
Since the numbers of stages in these sections are unequal, several The typical temperature proles in both column sections show
integrated congurations of HIDiCs are possible. As the number of that the location of minimum temperature driving force (DTi,min)
stages in the stripping section is lower than that in the rectifying appears at the bottom end of the stripping section, unlike the HIDiC
for benzene-toluene separation in which the DTi,min occurs around
the middle of the stripping section. With this characteristic, the
pressure ratio between column sections in propylene-propane
splitter cannot be reduced until the zero or negative temperature
Stage 1
Distillate
1
17
Feed
Stage 55 Uniform heat transfer area
16
TAC (million $/yr)

2 15
Compressor
Valve 3 14
Bottom
Stage 116 Uniform heat distribution
13

12
Stage 171 14 15 16 17 18
Prect (bar)

Fig. 10. The concentric HIDiC with middle-integrated conguration for PP splitter; the Fig. 12. Effect of rectifying pressure on the TAC of HIDiC with top-integrated cong-
number in circle represents the column section number for sizing and costing. uration and different heat distributing patterns for PP splitter.
1512 B. Suphanit / Energy 35 (2010) 15051514

21 Table 2
Optimum results of various HIDiCs for PP splitter.
20
Uniform heat transfer area Optimum parameters Top- Middle- Bottom-
19
integrated integrated integrated
TAC (million $/yr)

18 Prect (bar) Uniform A 14.72 14.12 13.25

Uniform Q 14.56 14.08 13.3
17
TAC (million $/yr) Uniform A 13.16 14.43 24.53
16 Uniform Q 12.67 14.31 26.18
Uniform heat distribution Power requirement Uniform A 6.87 7.43 13.52
15
(MW) Uniform Q 6.37 7.32 14.76
14 Total heat integration Uniform A 86.95 108.71 256.39
(MW) Uniform Q 83.37 108.20 276.02
13 Vapor through Uniform A 25,992 31,703 71,614
compressor (kmol/hr) Uniform Q 24,835 31,508 77,089
12
13 14 15 16 17 18
Prect (bar)

Fig. 13. Effect of rectifying pressure on the TAC of HIDiC with middle-integrated
larger energy penalty than the heat transfer around the top of the
conguration and different heat distributing patterns for PP splitter. rectifying section. This results in a large amount of total heat
transfer between sections, a large vapor ow at the bottom of the
rectifying section (ow through the compressor), and also a high
driving force at any place is reached. If the pressure ratio is too low, electricity cost in the compressor (see Table 2). The effect of energy
thus showing negative DTi at the bottom end of the stripping penalty from this bottom-integrated conguration is larger than
section, the bottom stages around this location could not be heat- the benet obtained from the potential pressure reduction. The TAC
integrated stages. An external hot utility supply and a reboiler are in every distributing scheme is therefore increased as the stripping
then unavoidable. The amount of heat transfer in HIDiC will only be section is integrated towards the bottom end of the rectifying
partial; thus the full potential of heat integration cannot be fullled. section.
For this reason, the operating condition giving rise to partial HIDiC When comparing both heat distributing schemes, it can be seen
is not considered in this case study. that the uniform heat distribution scheme gives a better cost
In the top-integrated conguration, any stage in the stripping performance than the uniform heat transfer area in the top-inte-
section is correspondingly matched with the same stage number grated arrangement. This means that distributing a large amount of
in the rectifying section. For the middle-integrated case, every stage heat away from the main reboiler or the bottom of the stripping
in the stripping section is integrated with its corresponding stage in section will result in a larger penalty than distributing the heat
the middle location of the rectifying section, i.e. stage no. 55-116. from the main condenser on the rectifying side. The optimum heat
Lastly, stage no. 110-171 in the rectifying section will be integrated distributing scheme for top-integrated arrangement is therefore
with the stripping section in the bottom-integrated conguration. the uniform heat distribution scheme in which a lower amount of
By performing the sensitivity analysis with the simulation heat load is distributed around the top of the rectifying section than
approach as outlined earlier, the TAC proles with respect to the that in the uniform heat transfer area scheme. The difference in the
rectifying pressure in each integrated conguration can be optimum conditions between these distributing schemes becomes
obtained as shown in Figs. 1214 for the top, middle, and bottom- smaller as the stripping section is integrated with the middle
integrated congurations, respectively. The optimum rectifying location of the rectifying section (see Table 2, and also Fig. 13),
pressure and the related parameters in each scheme are summa- where the energy penalty from allocating heat away from the top of
rized in Table 2. It can be seen that as changing the integrating rectifying section starts to show its impact. Finally, in the bottom-
location from top to bottom, the optimum Prect could be reduced integrated arrangement, the HIDiC with uniform heat transfer area
owing to a large temperature driving force available when inte- is better than that with uniform heat distribution since the energy
grating around the bottom of the rectifying section. The rectifying penalty caused by the heat distribution in the rectifying section
pressure can then be reduced further. The cost related to this dominates the system. For the uniform heat transfer area scheme,
potential pressure reduction should become lower. However, the lower amount of heat load is allocated at the bottom of the
heat transfer further away from the main condenser will incur

45
45 43
Uniform heat distribution 41
Rectifying section
40 39
Temperature (C)

Potential for pressure reduction

TAC (million $/yr)

37
35 35 Ti,min = 3 C
33
30
Uniform heat transfer area 31
Split off some stages.
29 Move them downward.
25 Stripping section
27

25
20 1 21 41 61 81 101 121 141 161
13 14 15 16 17 18 Stage No.
Prect (bar)
Fig. 15. Column temperature proles showing potential for the integrated structure
Fig. 14. Effect of rectifying pressure on the TAC of HIDiC with bottom-integrated with split stripping section evolving from the top-integrated structure in PP splitter
conguration and different heat distributing patterns for PP splitter. case.
 B. Suphanit / Energy 35 (2010) 15051514 1513

45 Table 4
Optimum results of various top-integrated HIDiCs with split stripping section and
43
uniform heat distribution for PP splitter.
41
Rectifying section No. of stages Distance shifted Optimum TAC
39 Ti,min = 3 C shifted downward (stages) Prect (bar) (million $/yr)
Temperature (C)

37 Feed Distillate 1 3 14.54 12.88

5 12 14.54 12.91
35 1
2 10 18 14.54 12.95
33 20 27 14.44 13.10
31 Valve
30 39 14.34 13.50
3
Compressor
Bottom 40 56 14.12 14.47
4
29
Stripping section
27

25
Then, for any selected number of stages at the bottom of the
1 21 41 61 81 101 121 141 161
Stage No. stripping section, the distance to be shifted could be determined
graphically by moving those selected stages downward (to the right
Fig. 16. An example of using column temperature proles to determine the shifted on the temperature-stage diagram) while simultaneously shifting
distance in the top-integrated structure with split stripping section for PP splitter case
down the rectifying temperature prole until the original value of
and the resulting concentric HIDiC structure.
DTi,min is obtained (3 and 2.5  C for uniform heat transfer area and
uniform heat distribution scheme respectively). An example of this
rectifying section when compared to the uniform heat distribution procedure is illustrated via Figs. 15 and 16 for top-integrated HIDiC
scheme. The energy penalty from the uniform heat transfer area with uniform heat transfer area. For any other number of stages
scheme is therefore lower in the bottom-integrated case. Most of being split off, the corresponding distance to be shifted away can
the optimum HIDiC structures for PP splitter can satisfy the column also be determined graphically. As illustrated in Fig. 16, the struc-
physical constraint, and thus the structures can be constructed as ture of this HIDiC now becomes more complicated. It consists of
concentric columns except the bottom-integrated structure with two concentric parts, one on top of the other which is larger in
uniform heat distribution. Consequently, under the conditions in diameter. Some piping works are necessary to connect between
this study, the most suitable HIDiC conguration for PP splitter is two split stripping sections. The column cost then requires the
the top-integrated concentric column with uniform heat distribu- estimation of 4 different column shells. This procedure is carried
tion operating at Prect of 14.56 bar. This optimum value of Prect is out with both distributing schemes, starting from their optimum
comparable to the result obtained in the previous study by Olujic top-integrated conditions. For each value of the number of stages
et al. [8] which suggested the Prect value of 14.6 bar. shifted, the value of Prect is varied via the sensitivity analysis to
However, as shown previously by Gadalla et al. [15], it may be determine the optimum condition. The results at some selected
attractive to explore other HIDiC arrangements, such as top-inte- cases are shown in Tables 3 and 4 for uniform heat transfer area and
grated structure with split stripping section. As noticed from the uniform heat distribution, respectively. From both distributing
shapes of the typical temperature proles of the PP splitter (Fig. 15), schemes in the top-integrated structure with SSS, it can be seen
the integration at a location away from the top of the rectifying that the cost performance does not improve at all when compared
section will be provided with a large driving force, hence a potential to their optimum top-integrated structures; although the rectifying
for rectifying pressure reduction. Therefore, the other integrated pressure could be reduced as the number of shifted stages is
structure to be further explored is the top-integrated structure with increased. It can be concluded that the location of heat integration
split stripping section (SSS). Some stages at the bottom of the in the rectifying section is crucial for obtaining the optimum HIDiC
stripping section will be moved downward and integrated with the conguration. The top-integrated arrangement with uniform heat
stages away from the original top-integrated structure. In this way, distribution is therefore the most suitable HIDiC design for PP
the potential for rectifying pressure reduction can be achieved. The splitter under the condition in this study.
trade-offs between the positive effect from potential pressure
reduction and the negative effect from heat transfer further away
from the top of the rectifying section can be evaluated. Exploring 4. Conclusions
this conguration will involve solving many possible scenarios
since the number of stages to be shifted away and the shifted Apart from the pressure ratio between the column sections, the
distance now become two additional variables which could heat distributing pattern is also another critical factor affecting the
produce a large number of possible design alternatives. Hence, only design and optimization of the HIDiC. The investigation of two
some alternatives are explored in order to see the trend of TAC. different heat distributing schemes; namely uniform heat transfer
First, starting from the optimum top-integrated arrangement, the area and uniform heat distribution, were thoroughly investigated
value of DTi,min at the bottom of the stripping section can be found. in this work. The study was carried out using a new approach to
simulate the HIDiC with maximum possible heat integration. It was
clearly demonstrated that the design and optimization of HIDiC
Table 3
Optimum results of various top-integrated HIDiCs with split stripping section and
became easier with this simulation model in which the heat
uniform heat transfer area for PP splitter. distributing pattern was set ahead of the simulation. The rectifying
pressure then became the only optimization variable to be
No. of stages Distance shifted Optimum TAC
shifted downward (stages) Prect (bar) (million $/yr)
considered, while all stage heat loads became tear variables which
could be changed under a particular distributing pattern during the
1 3 14.76 13.36
5 9 14.72 13.37 simulation. The results showed that the most suitable heat
10 14 14.66 13.36 distributing scheme was case-specic. For benzene-toluene sepa-
20 22 14.56 13.36 ration, the cost performance of HIDiC with uniform heat transfer
30 30 14.44 13.47
area was better than that of the equivalent structure with uniform
40 43 14.30 13.84
heat distribution. The optimum HIDiC structure for benzene-
1514 B. Suphanit / Energy 35 (2010) 15051514

toluene separation was the structure with uniform heat transfer R Reux ratio
area and one heat transfer stage on each end of the sections. The T Temperature
optimum Prect for this structure was at 1.88 bar. Despite the fact that U Overall heat transfer coefcient
it is impractical to construct this optimum HIDiC as a concentric
column due to the physical limitation of the column, the structure Subscripts:
of two separated columns with external exchanger could be applied i Stage no.
as an alternative. The TAC of the two-column structure was R, rect Rectifying section
approximately the same as the optimum TAC of the equivalent S Stripping section
concentric structure at Prect of 1.88 bar. In case of PP splitter, the T Total
effect of heat distribution on the HIDiC design and optimization min minimum
was more pronounced. Since the numbers of stages in both sections
were unequal, the stripping section could be integrated with the Greek letters
rectifying section at several possible locations. The location of heat D Difference
rejection along the rectifying section could affect the total amount
of heat integration between both sections. Heat rejection further References
away from the top of the rectifying section could result in high
energy penalty on the total heat integration between column [1] Olujic Z, Jodecke M, Shilkin A, Schuch G, Kaibel B. Equipment improvement
sections. A high vapor ow and a large power were then required as trends in distillation. Chem Eng Proc 2009;48:1089104.
[2] Petlyuk FB, Platonov VM, Slavinski DM. Thermodynamically optimal method
a consequence. Therefore, out of the three structures being for separating multicomponent mixtures. Int Chem Eng 1965;5(3):55561.
considered, the bottom-integrated structure showed the worst cost [3] Nakaiwa M, Huang K, Naito K, Endo A, Akiya T, Nakane T, et al. Parameter
performance, while the top-integrated was the best one. In the top- analysis and optimization of ideal heat-integrated distillation columns. Comp
Chem Eng 2001;25:73744.
integrated region, the inuence of heat distribution on the strip- [4] Mah RSH, Nicholas Jr JJ, Wodnik RB. Distillation with secondary refIux and
ping side was however larger than that on the rectifying section. vaporization: a comparative evaluation. AIChE J 1977;23(5):6518.
Thus, a larger amount of heat should not be assigned at the top of [5] Takamatsu T, Nakaiwa M, Huang K, Akiya T, Noda H, Nakanishi T, et al. Simu-
lation oriented development of a new heat integrated distillation column and its
the rectifying section (a long distance away from the bottom of the characteristics for energy saving. Comp Chem Eng 1997;21(Suppl. 1):S2437.
stripping section). The appropriate distribution scheme was the [6] Nakaiwa M, Huang K, Owa M, Akiya T, Nakane T, Sato M, et al. Potential energy
uniform heat distribution which allocated less amount of heat at savings in heat-integrated distillation column. Appl Therm Eng 1998;18:
107787.
the top of the rectifying section than the structure with uniform [7] Iwakabe K, Nakaiwa M, Huang K, Nakanishi T, Rsjorde A, Ohmori T, et al.
heat transfer area. The optimum Prect of the top-integrated HIDiC Energy saving in multicomponent separation using an internally heat-inte-
with uniform heat distribution was 14.56 bar. This value was very grated distillation column (HIDiC). Appl Therm Eng 2006;26:13628.
[8] Olujic Z, Sun L, de Rijke A, Jansens PJ. Conceptual design of an internally heat
close to the result obtained from the previous study (14.6 bar) by
integrated propylene-propane splitter. Energy 2006;31:308396.
Olujic et al. [8]. The other integrated structure, i.e. the top-inte- [9] Schmal JP, Vanderkooi HJ, de Rijke A, Olujic Z, Jansens PJ. Internal versus
grated with split stripping section, was also investigated for external heat integration operational and economic analysis. Chem Eng Res
possible performance improvement. However, this structure Des 2006;84(A5):37480.
[10] Nakaiwa M, Huang K, Owa M, Akiya T, Nakane T. Operating an ideal heat
showed no advantage over the basically top-integrated structure integrated distillation column with different control algorithms. Comp Chem
since the heat load allocation on the rectifying side still had a bigger Eng 1998;22(Suppl. 1):S38993.
impact than the potential pressure reduction from the split HIDiC [11] Nakaiwa M, Huang K, Naito K, Endo A, Owa M, Akiya T, et al. A new cong-
uration of ideal heat integrated distillation columns (HIDiC). Comp Chem Eng
structure. For both heat distributing schemes, none of these split 2000;24:23945.
structures performed better than the top-integrated structure in [12] Huang K, Shan L, Zhu Q, Qian J. A totally heat-integrated distillation column
terms of TAC. (THIDiC) the effect of feed pre-heating by distillate. Appl Therm Eng
2008;28:85664.
[13] Ho TJ, Huang CT, Lin JM, Lee LS. Dynamic simulation for internally heat-inte-
Acknowledgement grated distillation columns (HIDiC) for propylene propane system. Comp
Chem Eng 2009;33:1187201.
[14] Naito K, Nakaiwa M, Huang K, Endo A, Aso K, Nakanishi T, et al. Operation of
The author would like to thank Dr. Hong-Ming Ku for his useful
a bench-scale ideal heat integrated distillation column (HIDiC): an experi-
comment and suggestion. mental study. Comp Chem Eng 2000;24:4959.
[15] Gadalla M, Olujic Z, Sun L, de Rijke A, Jansens PJ. Pinch analysis-based
approach to conceptual design of internally heat-integrated distillation
Nomenclature
columns. Chem Eng Res Des 2005;83(A8):98793.
[16] Gadalla M, Jimenez L, Olujic Z, Jansens PJ. A thermo-hydraulic approach to
conceptual design of an internally heat-integrated distillation column
n Number of heat transfer stages (i-HIDiC). Comp Chem Eng 2007;31:134654.
[17] Douglas JM. Conceptual design of chemical processes. New York: McGraw-
P Pressure Hill; 1988.
Q Heat ow [18] Economic indicator. Chem Eng 2008;92.