You are on page 1of 7





Delft University of Technology, Laboratory for Process Equipment, Delft, The Netherlands
his paper demonstrates the usability of pinch analysis as a tool for conceptual design
of internally heat integrated distillations columns (HIDiCs). Incorporated appropri-
ately into the overall design procedure, pinch analysis enables a fast approach to an
optimum thermal performance, while bringing new insights and improving the understanding
of the nature of HIDiC designs. As illustrated by using a state of the art propylene splitter as
the base case, this approach emerged into an essential tool for the identication of economi-
cally interesting congurations for HIDiC applications.
Keywords: pinch analysis; HIDiC; heat integration; distillation; HEN.
The need for more sustainable developments in process
industries has encouraged further evaluations of the potential
for improving the energy efciency of the distillation
columns, which are, by far the largest energy consumer in
chemical process industries. This has led to a renewed inter-
est for the practical realization of an internally heat inte-
grated column, called HIDiC (Nakaiwa et al., 2003; Olujic
et al., 2003). This heat-coupling concept offers potential
for an ultimate saving in the energy consumption of single
columns. Namely, it combines the advantages of direct
vapour recompression at lowest feasible pressure ratio with
a nearly diabatic operation. Thus it is particularly suited as
an alternative for the established heat-pump assisted distilla-
tions, such as those encountered in polymer grade separ-
ations of close boiling mixtures; propylene-propane splitter
(PP-splitter) being the most prominent case of this sort.
According to the basic principle of the HIDiC, illustrated
schematically in Figure 1, heat is transferred from the hot
rectication section, operating on a higher pressure, into
the cold stripping section on a tray-to-tray basis. To
minimize the heat transfer effort, a concentric column
with rectication section placed within the stripping section
appeared to be the most suitable conguration for tray
columns (de Graauw et al., 2003), which dominate the
eld of the above mentioned applications.
The idealized design of HIDiC columns assumes an
equivalent number of thermally coupled stages of the
two-column sections, with a constant heat duty per stage
(Nakaiwa et al., 2003). However, this is not often the
case in practice. For instance, a PP-splitter usually contains
many more trays in the rectication than in the stripping
section, and as demonstrated by Sun et al. (2003) it
appeared that several coupling schemes are possible. The
most suitable option was found to be that with the stripping
section trays connected to the corresponding number of
trays in the upper part of the rectication section, which
allows the non-integrated part of the rectication section
to operate as a conventional column. Such congurations
will be called partial HIDiCs.
Since this case indicated rather a strong effect of the
coupling arrangement, it became obvious that for each
application an optimization study is necessary to arrive at
the best design. So, there is a necessity for a simple
design tool to guide the designer to arrive at a good
design and provide insights for improvement. Therefore,
an attempt was made in this paper to evaluate the suitability
of pinch analysis for this purpose, which over the last three
decades proved to be an outstanding conceptual tool to
improve the energy efciency of various process systems,
including distillation columns. Pinch technology concepts
have been widely applied to several studies and projects
for different design objectives including reducing operating
costs and energy consumptions. The literature is rich in
examples and successful applications for such a technique.
However, none of the published research on HIDiC designs
has considered this technique. In this paper, pinch analysis
principles and design guidelines are adopted and adapted
accordingly to provide guidance for conceptual design
and improvement of the energy efciency of HIDiCs.
Operation and controllability of HIDiCs are however not
discussed in this work. A separate study on these issues

Correspondence to: Dr M. Gadalla, Delft University of Technology

Laboratory for Process Equipment, Leeghwaterstraat 44, 2628 CA Delft,
The Netherlands.
E-mail:; Tel: +31(0)15 27 88901; Fax:
+31(0)15 27 86975.
# 2005 Institution of Chemical Engineers Trans IChemE, Part A, August 2005
doi: 10.1205/cherd.04301 Chemical Engineering Research and Design, 83(A8): 987993
will be devoted to this important subject. In present work,
pinch analysis is used for two separate tasks. First, to
design a given HIDiC as a heat exchanger network, this
will result in a minimum exchanger areas and energy
demands. In this case, HIDiC is preliminarily simulated
using any available commercial simulation package.
Then, process streams ow rate, temperatures and other
properties like specic heats are extracted to be used for
pinch analysis design. Basic data required in such a case
are feed specications and quantity, column data and separ-
ation specications, in addition to simulation output. The
second task is to apply the principles of pinch analysis to
re-examine and improve an existing design.
In various chemical plants such as reneries, networks
consisting of many exchanger units are used to heat process
feeds by recovering valuable heat from processes (e.g., dis-
tillation column, reactor, and so on). Pinch analysis proved
to be an efcient tool for designing such systems with
minimum utility consumptions and capital costs (Asante
and Zhu, 1997; Gadalla et al., 2003). A heat exchanger net-
work (HEN) produced by pinch techniques is characterized
by minimum exchanger areas, optimum distribution of heat
loads on exchanger units and minimum hot and cold utility
consumptions. A HIDiC design can be performed in a new
method in which the column conguration is represented as
a heat exchanger network. The rectifying column will be
seen as a large heat source, and similarly the stripping
column will be considered as a large heat sink. From this
point of view, the HIDiC can be designed as an exchanger
network that recovers heat from the rectifying section
and provides it to the stripping section. The pinch concept
from the point of view of HIDiCs is represented in
Figure 2. The rectifying column is represented by a hot
stream using a cooler which represents the condenser of
the column. On the other hand, the stripping column is
indicated by a cold stream that exchanges heat with the rec-
tifying hot stream through a number of exchangers. Part of
the heat required by the stripping column is provided using
an external heater, i.e., the reboiler of the column. The heat
exchangers of the network are the heat panels installed on
the column stages. Theoretically, the number of exchanger
units is equivalent to the number of stages being thermally
The required data for design are the stage liquid and
vapour stream ows and temperatures of both columns,
and similar details for the condenser and reboiler streams,
in addition to the enthalpy change of each process
stream. The temperatureenthalpy diagram is a useful
representation of the thermal data of process streams.
Each process stream can be represented by a straight line
(sometimes a curve) on the temperatureenthalpy graph.
All hot streams are grouped as one big hot stream, called
hot composite curve. Similarly, all cold streams are
combined into a cold composite curve. These curves are
called composite curves of a process. According to
Linnhoff et al. (1982), composite curves represent only
the product streams and utility streams. The two curves
can be plotted together on the same graph by separating
them by a certain temperature difference, which is known
as the minimum temperature difference. This is illustrated
in Figure 3, which shows new composite curves con-
structed for a partial HIDiC design. The upper line
represents the hot streams, i.e., the rectifying stages,
while the lower line is for the cold streams or the striping
stages. This gure includes all the information of the exter-
nal and internal streams and the column stages, unlike the
conventional composite curves which account only for
the external process streams. The overlap area between
the hot curve and cold curve indicates the maximum
amount of heat recovery for a given temperature difference.
The upper horizontal (x-axis) difference between the two
curves is equivalent to the minimum reboiler duty, while
the lower horizontal difference gives the minimum
Figure 1. Operating principle of a HIDiC.
Figure 2. A HIDiC represented as a heat exchanger network.
Figure 3. HIDiC composite curves.
Trans IChemE, Part A, Chemical Engineering Research and Design, 2005, 83(A8): 987993
988 GADALLA et al.
condenser duty. When HIDiC is ideal, the two curves will
start at the same vertical point lines, i.e., the reboiler duty
and/or the condenser duty become zero. In fact, in this
case, the reboiler and/or the condenser unit are not
needed. Design of HIDiCs as a heat exchanger network
can then be made using pinch analysis principles (Linnhoff
et al., 1982). The design requires an assumption to be made
on the minimum allowed temperature difference. Also,
some constraints need to be imposed, like stage matches,
maximum physical space and heat duties. The new rep-
resentation of HIDiC (Figures 2 and 3) provides insights
into the nature of thermal performance of HIDiCs and, as
it will be shown later, used as a conceptual tool, it enables
arriving at minimum total cost design.
Composite curves are constructed normally for
distillation columns for the external process streams, i.e.,
feed, products, and condenser and reboiler streams.
Linnhoff et al. (1982) provides a calculation procedure in
order to obtain composite curves for a given process.
However, this representation has little use for HIDiC con-
gurations because it only considers the external streams.
For HIDiC, it is recommended to represent the internal
vapour and liquid streams on all thermally coupled
column stages. This provides a full picture of the HIDiC
process and may lead to valuable insights into the nature
of the process. Figure 4 shows typical composite curves
for a HIDiC design for the separation of a benzenetoluene
mixture. Necessary data for this hypothetical case are con-
tained in Figure 4. To obtain such a plot shown in Figure 4,
distillation column is rst simulated using ASPEN

. A
detailed illustration of the simulation and design of
HIDiCs can be found elsewhere (Gadalla et al., 2004).
Then, the obtained liquid and vapour stream temperatures
and enthalpy changes are extracted and imported to
ASPEN pinch to plot the process composite curves.
These curves include the information on temperatures,
ow rates and specic heats of all process streams, such
as condenser, reboiler and stage vapour and liquid streams.
One may notice that the HIDiC design of Figure 4 is an
ideal one since the reboiler duty is equal to zero. This can
be seen on the right side of this gure when the two curves
start on the same vertical enthalpy line. As shown, the con-
denser duty is equivalent to 413 kW and the amount of heat
recovery, which is indicated by the overlap between the
two curves, is 2866 kW. The pinch point can be represented
by a vertical line in the composite curve diagram. This line
cuts the composite curves at the minimum temperature
difference (separation distance). The temperatures of the
pinch are those at which the composite curves are cut by
the pinch line. The pinch divides the design of the process
into two stages, and hence classies the streams into two
groups, streams above the pinch and streams below the
pinch. The main principles of the design using pinch analy-
sis are: no cooling is allowed above the pinch and similarly
no heating below the pinch. In addition, no heat is trans-
ferred across the pinch. Satisfying these guidelines will
reduce the energy consumption of the distillation processes
(Smith, 1995).
To incorporate more information of the distillation
process to the composite curves, the heat transfer processes
across the column stages can be superimposed. As shown in
Figure 4, this can be done by representing the heat transfer
on each stage by a straight line connecting the temperature
of a rectifying stage with its corresponding stripping stage.
Each line represents the process of heat transfer from one
stage in the rectifying column into the corresponding stage
in the stripping column. These lines are equivalent to
exchanger units or heat panels placed on the column
stages. Each line starts on the hot composite curve at the
temperature of the hot vapour and ends on the cold compo-
site curve at the temperature of the cold liquid.
The height of a line is equivalent to the temperature driv-
ing force across the stage through which heat is transferred;
however, the width does not necessary indicate the duty of
this heat transfer. As shown, some lines cross over the
pinch; this means that these lines transfer heat from
above the pinch to below the pinch, which is a violation
of the principles of pinch analysis. This across-pinch heat
transfer is conducted on the stages 4, 5 and 6. This may
incur more energy consumption, i.e., the reboiler duty or
condenser duty is not at the minimum value. As the reboiler
duty is already zero for this ideal HIDiC design, the con-
denser duty (413 kW) must be at higher value than the
minimum one. Consequently, the amount of heat recovery
is higher than what is necessary to accomplish an ideal
HIDiC. In other words, it may be expected to reach the
same HIDiC design with smaller heat transfer areas and
less utility consumption. Thus, to improve the performance
of the original design, the circumstances that violate the
pinch technology rules should be avoided. As a result,
heat transfer across these three stages should not be per-
formed. This calculation is performed using ASPEN

a way that the heat transfer across these pinched stages is
not considered. This will lead to a new heat integration
scheme with an opportunity for better energy savings.
The modied heat integration scheme will then have heat
transfer throughout all column stages except stages 4, 5, 6
(Figure 5). The modied design in practice will have heat
transfer panels only on the integrated stages, whereas
Figure 4. HIDiC composite curves for benzenetoluene separation
(100 kmol/h equimolar feed; 20 stages; pressure ratio (bar): 2.5:1.0;
stage duty: 287 kW; 99.5% benzene in distillate and 0.5% in bottoms,
including stages with heat transfer operation.
Trans IChemE, Part A, Chemical Engineering Research and Design, 2005, 83(A8): 987993
stages 46 will operate without panels. And since the
location of the pinch point is only sensitive to the tempera-
ture proles which in turn are governed by the pressure
ratio, the location of the pinched stages is then stable.
In Table 1, the results of the modied design are com-
pared with the original design. These results as mentioned
previously were obtained from ASPEN

simulation tool.
The total annual cost reported in the table includes the
investment costs of column, condenser, compressor and
heat transfer mediums (devices). Reboiler cost is not
contained since HIDiC is ideal and reboiler is not needed.
The cost also accounts for the electricity and water
consumption. Cost correlations for equipments including
column shell and trays, compressor, condensers, reboilers
and heat transfer devices and operating costs of electricity
and cooling water are those proposed by Douglas (1998).
To account for the complexity of HIDiC compared to
normal distillation columns, the installed column cost is
multiplied by 1.5. One may notice that the comparison of
Table 1 does not include any control or startup issues
since the study is devoted to conceptual design only.
Since here only two ideal options are considered, the
exclusions of reboiler and condenser, required to startup,
is not essential, because the costs related to these units
are the same for both cases. As shown in Table 1, the
modied design requires less heat to achieve an ideal
HIDiC. The condenser duty and the water consumption
are both reduced and the total heat transfer area required
is 36% less than that needed for the original design. The
compressor load saving is about 4% and hence its electri-
city consumption will also reduce. The modied design
has a lower total cost with saving of 4% with respect to
the original design cost.
From Figure 6, it can be seen that thanks to a substantial
reduction in the total heat transfer area the area required per
stage will remain relatively small. Such an impressive gain
on the total heat transfer area side has arisen from the fact
that in case of modied design the temperature driving
force increased substantially (see Figure 7), due to avoiding
the heat transfer mismatches. The application of these new
ideas may lead to considerable savings in cases where the
condenser duties are larger than reboiler duties. In these
cases, additional savings in hot utility consumptions
could be attained.
In practice the applications amenable for the implemen-
tation of this heat integration concept are those where a
heat pump is already established as a technology, i.e., in
energy intensive separations of close boiling mixtures
such as propylenepropane and the like separations.
Due to different product purity requirements, a propylene
propane separation column (so called PP-splitter) contains
usually much more stages in the rectifying than in the strip-
ping section of the column. It was observed by a separate
study done on HIDiCs by Sun et al. (2003) that a top
design is much more efcient than a bottom alternative
for a propylenepropane splitter. In the top design, the
stripping stages are integrated with the equivalent number
of the top stages of the rectifying section, while the lower
(non-integrated) part of the rectifying section operates as
a normal column. The opposite is the case with the
bottom design.
The relative performance of the two cases is illustrated in
Table 2, which includes a schematic illustration of the top
Table 1. Results of a modied HIDiC design for benzenetoluene
Parameter Original design Modied design
Reboiler duty (kW) 0 0
Stage heat duty (kW) 287 391
Compressor load (kW) 446 427
Heat recovery (kW) 2866 2737
Condenser duty (kW) 413 389
Water utility (t/y) 79.040 74.448
Total heat transfer area (m
) 363 233
Total annul cost (k$/y) 523 503
Figure 6. Heat transfer areas of the modied HIDiC and original designs.
Figure 5. A new heat integration scheme for an optimum HIDiC design.
Figure 7. Temperature driving forces of the modied HIDiC and original
Trans IChemE, Part A, Chemical Engineering Research and Design, 2005, 83(A8): 987993
990 GADALLA et al.
and bottom designs. The corresponding operating
conditions, for the base case HIDiC-type PP-splitter, are
given in Table 3.
Obviously, all the duties and the corresponding heat
transfer areas are much larger in the case of bottom
design. Here, the upper, normally operating part of recti-
cation section does not perform adequately, which means
that practically all separation work is concentrated in a
rather short integrated part, where an immense reux/boil
up is required to perform accordingly with a relatively
small number of theoretical plates. Namely, the reux
ratio in the upper (normal part) is well below the minimum
for normal operation (this part of the section is far from the
feed tray and thus it can be operated at such a theoretically
unsound condition). Although this conguration is not
worth of further elaboration, it is evaluated here together
with the most benecial one by using the stage temperature
proles and composite curves, because this provides useful
insights into the conguration imposed nature of the heat
transfer process.
Figure 8 shows the temperature of the column stages for
each design before transferring heat from the rectifying
column to the stripping column as well as after the heat
is being transferred between the two columns. Strikingly,
in the case of the bottom design, the rst 17 stripping
stages are cooled down after transferring heat from the
rectifying column. This means that the cold liquid streams
on these stages are losing heat and becoming colder. This
indicates that these streams consume cold utilities as hot
streams rather than receiving heat from other hot streams.
This is a consequence of the fact that due to insufcient
separation in the rectication section the top stages of
stripping section contain relatively much more propylene
and consequently have a lower temperature at the same
pressure. The only remedy would be increasing the com-
pression ratio accordingly, which however is detrimental
for compression duty. On the other hand, the lowest
feasible compression ratio for HIDiC is that that ensures
that the lowest temperature in the rectication section is
higher than the highest temperature in the stripping section.
Certainly, minimum temperature difference implies maxi-
mum heat transfer area and this is a typical trade off con-
sideration during design of installations for heat transfer.
Returning to the above discussed ill-situation, in overall,
the amount of heat recovered from the rectifying stages is
less than what is available and on the contrary, excess cool-
ing utilities are consumed. As a result, more heat needs to
be transferred between the two columns in order to reduce
the reboiler duty further. On the other hand, all the strip-
ping stages in of heat recovery available for this case is
completely exploited in heating the cold streams of the
top design are heated by recovering heat from the rectifying
stages. In addition, the amount the stripping stages and
hence similar situations are avoided.
Figures 9 and 10 show the grand composite curves for
the top and bottom designs respectively. Grand composite
curves are another form of the temperatureenthalpy
representation of a given process (Smith, 1995). In such
case, a line with a negative slope represents a hot stream
or a group of streams, while cold streams are represented
by a line with a positive slope. A hot stream can be used
to heat a cold stream if there is a viable temperature driving
force. The gap area between the two hot and cold lines is
equivalent to the amount of heat recovery. Additionally,
the pinch point can be detected from the grand composite
curve; it is located where the curve touches the zero-
Table 2. Results of HIDiC top and bottom designs for propylenepropane
Stage duty (kW) 1817 4686
Condenser duty (kW) 5782 18 880
Reboiler duty (kW) 0 0
Compressor duty (kW) 7489 18 820
Total transfer area (m
) 25 364 37 360
Table 3. Problem data and HIDiC design specications.
Column specication
Feed propylenepropane
Composition propylene mole% 50
Flow rate t/h 111.6
Pressure bar 12.2
Temperature 8C 31.7
Rectifying pressure bar 19.2
Stripping pressure bar 12.2
Rectifying stages 154
Stripping stages 57
Top product purity propylene mole% 99.6
Bottom product purity propylene mole% 1.1
Figure 8. Stage temperature proles for top and bottom HIDiC designs.
Figure 9. A grand composite curve of the top HIDiC design.
Trans IChemE, Part A, Chemical Engineering Research and Design, 2005, 83(A8): 987993
enthalpy axis. The grand composite curve of the top
design shows that all the rectifying section hot streams
can provide heat to the stripping section cold streams.
Therefore, all heats available in the rectifying section are
transferred into the stripping section. The cooling duty
required, as shown in the Figure 9, represents only the
necessary cooling duty of the condenser. Conversely, the
grand composite curve of the bottom design reveals that
a part of stripping stage streams behaves as hot streams;
these streams were supposed to be integrated with the
rectifying hot streams. Alternatively, they are cooled,
consuming additional cooling duties and reducing the
actual amount of heat recovery from the rectifying stages.
Moreover, the amount of heat recovered is shown to be
less than which is available on the rectifying stages. In
conclusion, the HIDiC bottom design requires more
heat transfer in order to compensate the shortcoming of
the design performance, and consequently, poor energy
efciency is obtained.
According to Figures 8 and 9, the pinch point situation
occurs on the bottom stages of the HIDiC top design.
This indicates that it could be possible to improve the
design by changing the heat integration scheme. An attempt
for improving the energy performance of the top design
delivered two possible congurations, i.e., two modi-
cations of the original design shown schematically in
Figure 11. The rst improvement is a change in the heat
transfer rate across the rst few top and the last few
bottom stages of the column. In this design modication
(top-new), the heat duty is increased on these stages.
The design transfers a 2600 kW of heat across the stages
14 and 5457 and all others exchange a constant heat
duty of 1600 kW. These heat duty values are obtained by
using ASPEN

simulation in such a way that the reboiler

duty is reduced to reach zero. This design exploits the
large temperature driving force of the top stages and as a
result, less total heat needs to be integrated to achieve
zero reboiler duty (ideal), compared to the original design
(see Table 4). Also, the compressor doesnt need to com-
press so much vapour and similarly the condenser is not
forced to condense additional amounts of vapours. So,
The advantages of the top-new design is that less heat
transfer area is required on most of the column stages
and only few stages at the top and bottom need larger
amounts. In overall, the electricity and cooling water oper-
ation costs are less than those needed by the original
design. In addition, the equipment investment costs of com-
pressor, condenser and heat transfer panels are relatively
cheaper. The design has a total cost saving of 3% with
respect to the original design. The control challenges of
this design are not different from those for the original
top design since both congurations adopt the same
heat integration scheme.
Alternatively, the bottom stripping stages are integrated
with some stages at the bottom of the rectifying section
(top-bottom). This new design transfers a constant heat
rate of 1790 kW across all column stages; however, the
last three stages (5557) of the stripping column are inte-
grated with the stages 9799 of the rectifying column.
The total heat integration requirement is still smaller than
that required for the original design. As result, the top-
bottom design has the advantages of requiring a relatively
smaller and constant amount of heat transfer area on all
integrated stages compared to the original top design.
Also, the compressor and condenser loads are relatively
smaller. However building such a two-section HIDiC
conguration may not be feasible because of the related
mechanical design complexities.
According to the results summarized in Table 4, both
modications result in a better performance. In general,
the top-new design requires less compressor load and
condenser duty as well as a smaller heat transfer area with
respect to the other alternatives. As a consequence, there is
a reduction in the total annual cost of the top-new design of
240 k$/y to the original design cost. These cost savings may
be of greater importance for cold-box distillations where the
refrigerant costs play an important role. This reveals that by
changing the heat integration scheme of a given design,
opportunities for further improvements may arise.
Pinch analysis principles have been incorporated into the
conceptual design of internally heat-integrated distillation
columns. A new representation of the HIDiC congurations
has been introduced, which allows both the design of an
optimum (minimum total costs) HIDiC and a better
Figure 10. A grand composite curve of the bottom HIDiC design.
Figure 11. Improvements to a HIDiC top design for propylenepropane
splitter (numbers refer to stage heat duties).
Table 4. Results of modied designs to a HIDiC top.
Parameter Original design Top-new Top-bottom
Reboiler duty (kW) 0 0 0
Compressor load (kW) 7489 7240 7383
Condenser duty (kW) 5782 5647 5777
Total heat transfer area (m
) 25 364 23 904 24 437
Total annul cost (M$/y) 8.54 8.30 8.44
Total heat integration (MW) 103.6 99.2 102.0
Trans IChemE, Part A, Chemical Engineering Research and Design, 2005, 83(A8): 987993
992 GADALLA et al.
understanding of the performance of such a complex
column. Pinch technology guidelines have been applied
successfully to improve the thermal performance of given
PP-splitter HIDiC designs.
Asante, N.D.K. and Zhu, X.X., 1997, An automated and interactive
approach for heat exchanger network retrot, Trans IChemE, Part A,
Chem Eng Res Des, 75(3): 349360.
Gadalla, M., Jobson, M. and Smith, R., 2003, Optimization of heat-
integrated renery distillation systems, Trans IChemE, Part A, Chem
Eng Res Des, 81: 147152.
Gadalla, M., Olujic, Z., de Rijke, A. and Jansens, P.J., 2004, A thermo-
hydraulic approach to conceptual design of an internally heat-integrated
distillation column (i-HIDiC), European Symposium on Computer-
Aided Process Engineering, 14: 181186.
de Graauw, J., Steenbakker, M.J., de Rijke, A., Olujic, Z. and Jansens, P.J.,
2003, Heat Integrated Distillation Column, WO 03/061802 A21.
Douglas, J.M., 1998, Conceptual Design of Chemical Process (McGraw-
Hill, New York, USA).
Linnhoff, B., Townsend, D.W., Boland, D., Hewitt, G.F., Thomas, B.E.A.,
Guy, A.R. and Marsland, R.H., 1982, A User Guide on Process
Integration for the Efcient Use of Energy (Institution of Chemical
Engineers, Rugby, UK).
Nakaiwa, M., Huang, K., Endo, A., Ohmori, T., Akiya, T. and Takamatsu, T.,
2003, Internally heat integrated distillation columns: a review, Trans
IChemE, Part A, Chem Eng Res Des, 81: 162177.
Olujic, Z., Fakhri, F., de Rijke, A., de Graauw, J. and Jansens, P.J., 2003,
Internal heat integrationthe key to an energy conserving distillation
column, J Chem Technol Biotechnol, 78: 241248.
Smith, R., 1995, Chemical Process Design (McGraw-Hill, New York,
Sun, L., de Rijke, A., Olujic, Z. and Jansens, P.J., 2003, Industrially
viable conguration for a heat integrated distillation column, The 5th
international Conference on Process Intensication for the Chemical
Industry, Maastricht, The Netherlands, 1315 October, BHR Group
2003, M. Gough (ed.), 151166.
The authors are thankful to the Program Ofce on Economy, Ecology
and Technology (Grant EETK01025) for nancial support and to the
project partners ABB-Lummus, AKZO-Nobel, BP, DSM, ECN, SHELL
GS and Sulzer Chemtech for their interest and the permission to publish
this work.
The manuscript was received 3 November 2004 and accepted for
publication after revision 7 June 2005.
Trans IChemE, Part A, Chemical Engineering Research and Design, 2005, 83(A8): 987993