Sie sind auf Seite 1von 13

ARTICLE

pubs.acs.org/IECR

Designing a Packed Dividing Wall Column for an Aromatics


Processing Plant
 arko Olujic*,
Igor Dejanovic, Ljubica Matijasevic, Helmut Jansen, and Z

Faculty of Chemical Engineering and Technology, University of Zagreb, Savska cesta 16, HR-10000 Zagreb, Croatia
Julius Montz GmbH, Postbox 530, 40705 Hilden, Germany

Process and Energy Laboratory, Delft University of Technology, Leegwaterstraat 44, 2628 CA Delft, The Netherlands

ABSTRACT: This paper introduces a comprehensive design method assembled using facilities of a commercial software package
that complemented by Excel programs, which contain own column dimensioning and well established cost estimation procedures,
enables proper assessment of the industrial viability of a dividing wall column (DWC) equipped with corrugated sheet structured
packings. The heart of the performance simulation tool is a detailed four-column model that in conjunction with a simple,
theoretically founded short-cut method providing reliable initial values for liquid and vapor splits and a simple but eective objective
function for design optimality indication allows determination of the adequate stage and reux requirement of a DWC. The
proposed dimensioning method enables a close approach in accuracy to that required at the stage of conceptual design for purposes
of making a bid by an equipment manufacturer. Compared to a two-columns-in-series conguration, as employed in an aromatics
processing complex within a renery, a DWC equipped with state-of-the-art structured packing and auxiliary internals requires
approximately 43% less energy to deliver three fractions at required product specications. This, accompanied by savings of nearly
51% based on total annualized costs, indicates that implementing a DWC could lead to a signicant increase in protability of
aromatics processing plants.

1. INTRODUCTION
Being the most widely used and most energy intensive among
large scale separation techniques, distillation became the main
target of eorts oriented toward increasing the sustainability of
process industries.1 However, the implementation of energysaving solutions is often capital intensive and process industries
are generally reluctant to implement them if this is not associated
with signicant improvement in the protability of a plant. This is
particularly the case in the petroleum rening world. As elaborated in greater detail by Hartman et al.,2 catalytic reforming to
process aromatics is an important economic factor for modern
reneries. Reactor euent stream is rich in benzene, toluene,
ethylbenzene, xylenes, and heavier aromatics, and an aromatics
complex usually contains several distillation columns arranged in
trains (sequences) to recover and separate the aromatic components into individual products and/or certain component-rich
fractions. A comprehensive review of the dividing wall column
(DWC) state of the art, including a survey of application-related
patents, indicated that aromatics complexes oer various opportunities for implementation of a DWC.3 The present paper is
concerned with design of a conventional, three-product DWC
suitable for a specic aromatics complex situation as encountered
in a Croatian renery.
The energy-saving potential in this type as well as in many
other applications is signicant and can be estimated with
condence using dierent simulation methods. The papers by
Tiantafyllou and Smith4 and Segovia-Hernandez et al.5 provide
some quantication in this respect, and the obtained results agree
well with the numbers reported for some real industrial applications in a paper by Kaibel et al.6
However, as indicated by Dejanovic et al.,3 and in a recent
state-of-the-art paper by Asprion and Kaibel,7 the columns
r 2011 American Chemical Society

dimensioning procedures still belong as proprietary knowledge


to a few equipment manufacturers active in this eld. In order to
arrive at total annualized costs, to enable comparisons of alternatives, some dimensioning-related eorts have been undertaken
in academic publications, e.g., refs 8 and 9. However, the nature
of applied approximations/simplications is such that it may lead
to erroneous conclusions on both the process design side and the
economics side.
One should realize that hydraulic design of the partitioned part
of a DWC is a delicate activity, and that pressure drop on two
sides of the wall must be equal. If not properly arranged in the
design phase, by adjusting the necessary amount of ow resistance exhibited by internals used in conjunction with xed
specic liquid ows, the equalization of the pressure will be
imposed by nature, i.e., by spontaneous adaptation of vapor ows
in two sections. This will inevitably lead to establishing operating
liquid to vapor ow ratios that could dier from that required to
achieve the desired degree of separation in beds on both sides of
the partition wall. This most distinctive feature of hydraulic
design of a DWC has not received adequate attention in the open
literature so far.
An objective of this paper is to ll this gap, and as will be
demonstrated later on, the method proposed in this paper allows
a close approach to actual design practices, as adopted by the
equipment manufacturing company Julius Montz, Hilden,
Germany, the pioneer in the eld of design and construction of
packed DWCs.
Received: October 5, 2010
Accepted: February 24, 2011
Revised:
January 20, 2011
Published: March 24, 2011
5680

dx.doi.org/10.1021/ie1020206 | Ind. Eng. Chem. Res. 2011, 50, 56805692

Industrial & Engineering Chemistry Research

ARTICLE

Table 1. Base Case Feed and Product Compositions


stream name
feed

C5C6

BRC

heavies

31.74

6.94

3.70

21.10

n-butane

0.019

0.088

0.000

0.000

isopentane

0.064

0.291

0.000

0.000

n-pentane

0.045

0.206

0.000

0.000

1

total ow [t h ]
component mass fractions

Figure 1. Conventional, two-columns-in-series conguration (indirect


sequence) for separation of a three-component or multicomponent feed
into three specied products or fractions, respectively.

With proper column shell dimensions and internal conguration, it is possible to obtain reliable installed equipment cost
estimates and consequently arrive at total annualized costs
(TACs) with enough condence to allow a realistic comparison
with established, two-columns-in-series congurations, i.e., a
proper assessment of potential benets of implementing a
DWC. As shown in what follows, the results of this study clearly
indicate that adopting a DWC as the standard design for
aromatics plant applications could lead to increased protability
of complex reneries.

2. DESIGN CASE
Upon a recent decision to concentrate mainly on fuel production, the aromatics complex at the INA (presently MOL Group)
renery in Sisak, Croatia, has been reduced to a minimum, i.e., to
a direct separation sequence containing two columnsthe
so-called platformate splitter and the benzene recovery
columnconnected in series (see Figure 1) to separate the
platformer reactor euent stream which contains some 40
components into three fractions: (1) C5C6 gasoline containing no more than 1.5 mass % benzene, (2) a benzene-rich cut
(BRC) containing 68 mass % benzene, and (3) a heavy reformate
stream (heavies) containing toluene, ethylbenzene, xylenes, and
heavier components with no more than 0.5 mass % benzene.
For purposes of this study the 40 components contained in the
actual feed stream have been lumped together into a representative 15-component mixture. The mass ow rates and compositions of the feed and product streams as considered in the present
simulation study are shown in Table 1.
In order to provide an appropriate basis for the evaluation and
comparison of related costs, two conventional columns are
considered here as new designs, and to provide complete
information both options, i.e., tray and packed columns, are
considered.
3. STAGE AND REFLUX REQUIREMENTS
The two columns of the base-case conguration (see Figure 1)
have been simulated using detailed methods available in ChemCAD. Regarding the simulation approach, a DWC, shown
schematically in Figure 2, has signicantly more degrees of

2-methylpentane

0.080

0.351

0.026

0.000

n-hexane

0.043

0.050

0.270

0.000

benzene
3-methylhexane

0.086
0.020

0.013
0.000

0.680
0.024

0.005
0.026

toluene

0.247

0.000

0.000

0.373

ethylbenzene

0.035

0.000

0.000

0.053

p-xylene

0.042

0.000

0.000

0.064

m-xylene

0.122

0.000

0.000

0.183

o-xylene

0.055

0.000

0.000

0.083

m-ethyltoluene

0.047

0.000

0.000

0.071

1,3,5-trimethylbenzene
1,4-diethylbenzene

0.077
0.017

0.000
0.000

0.000
0.000

0.116
0.025

Figure 2. Schematic representation of a DWC with indication of design


parameters.

freedom than a conventional distillation column. For instance,


the number of stages for six sections needs to be provided.
Additional parameters indicated as circles in Figure 2, specic to
the internal conguration as encountered in a DWC, are liquid
and vapor splits above and below the partition wall, respectively.
A good set of initialization data is essential to ensure convergence
5681

dx.doi.org/10.1021/ie1020206 |Ind. Eng. Chem. Res. 2011, 50, 56805692

Industrial & Engineering Chemistry Research

ARTICLE

Figure 4. Plot indicating optimum number of stages for a DWC.

Table 2. Base Case and DWC Simulation Results


base case
C1

Figure 3. Four-column model used to simulate the operation of a threeproduct DWC.

of the detailed simulation method. Fortunately, these can be


easily obtained using a rather simple but in this respect very
eective short-cut method, introduced recently by Halvorsen10
and Halvorsen and Skogestad.11,12
Following suggestions of Halvorsen and Skogestad,9 the
so-called Vmin diagram method has been implemented in ChemCAD. Since the original method requires dealing with pure
components, the 15-component feed has been represented by
three representative key components: 2-methylpentane,
benzene, and toluene. Then, the Vmin diagram was constructed
by rigorous simulation of possible binary splits, with the stage
number being at least 4 times the minimum number of stages,
4Nmin, and mass recoveries of key components set to 0.999.
From that diagram, appropriate estimates of required design
parameters for initialization of a detailed simulation were
obtained. Details related to Vmin diagram application in conjunction with a rigorous simulation model and corresponding
numbers can be found elsewhere, e.g., Dejanovic et al.13
Detailed simulations have been performed using the so-called
four-column model, shown schematically in Figure 3. This
conguration was the preferred choice in this case, because the
prefractionator and main column sides of the partitioned part of
the column are represented by separate column sections, which
makes the dimensioning eort more straightforward. Another
reason is that this is the only conguration that can be used for
dynamic simulation with the purpose of the control system study.
A detailed elaboration of relative advantages/disadvantages of

C2

C1 C2

DWC

number of stages

40

38

78

86

reux ratio

1. 70

2.39

2.80

reboiler duty (MW)


condenser duty (MW)

3.55
3.16

2.63
2.51

6.18
5.67

3.50
2.76

four dierent congurations that can be used to simulate the


performance of a three-product DWC can be found elsewhere,
e.g., Dejanovic et al.3
The rst step was to restore feed composition to the full 15
components. Benzene mass fractions in the distillate and the side
draw liquid ow rate were set to be the same as in the base-case
simulation, while the initial value for reboiler duty was set to give
the required minimum vapor ow. Liquid and vapor split ratios
were adjusted until the minimum possible mass fraction of
benzene in the bottoms was achieved, which was the same as
in the base case.
The next step was to determine the actual number of stages in
each section. This was done by reducing the number of stages in
sections keeping the mass fractions of benzene in the distillate
and bottoms, as well as liquid side draw ow rate, constant. For
every converged case, an optimization was performed using the
optimization tool built in ChemCAD, with the objective function
being min Qr, and the independent variables the liquid and vapor
split ratios.
To arrive at the optimum combination of reux ratio and the
number of stages, the empirical objective function, min N(R 1),
which approximates eectively the total annualized cost of a conventional distillation column, was used. According to Figure 4, the
minimum value corresponds to 64 equilibrium stages, based on the
main column stage count. The same result is obtained using all
equilibrium stages contained in the DWC.
In the present case the number of stages at the prefractionator
side and the number of stages at the main column side are equal,
i.e., 22. This means that the total number of stages contained in
the DWC is 86, which means eight stages more than required in
two sieve tray columns connected in series. The number of
equilibrium stages of compared congurations and the corresponding reux ratios are given in Table 2. The simulation results
for conventional, two-columns-in-series sequence and a DWC
5682

dx.doi.org/10.1021/ie1020206 |Ind. Eng. Chem. Res. 2011, 50, 56805692

Industrial & Engineering Chemistry Research


shown in Table 2 indicate that a DWC would require approximately 43.3% less energy, based on reboiler duties.
Figure 5 shows the algorithm for stage and reux requirement
calculations associated with a three-product DWC.

4. COLUMN DIMENSIONS
Upon completion of the performance calculations, the number of equilibrium stages (theoretical plates) in each column
section was xed as well as corresponding liquid and vapor ow
rates. For the tray column option, common cross-ow sieve trays
have been chosen with a tray spacing of 0.6 m. Using tray
eciencies as experienced in similar applications, the platformate
splitter contains 61 trays and the benzene recovery column
contains 59 trays, with the feed stage in both cases in the middle
of the column. For packed conventional columns and packed
DWC, Montz-pak B1-350 MN, a state-of-the-art, high performance corrugated sheet structured packing was chosen. Bed
heights in conventional and partitioned parts have been determined in accordance with the number of contained stages,
assuming an HETP value of 0.4 m. This number including the

Figure 5. Schematic illustration of stage and reux requirement calculation procedure.

ARTICLE

common safety margin is based on total reux test data obtained


with Montz-pak B1-350MN using the chlorobenzene/ethylbenzene system at atmospheric pressure.14
The number of packed beds to be installed in rectication and
stripping sections of a packed column has been determined based
on a rule of thumb indicating that the single bed height should
not exceed that equivalent to 20 equilibrium stages. If more than
20 stages are required in a section, then two beds should be
considered, which implies providing additional column height to
accommodate the necessary liquid redistribution section. The
latter consist usually of a liquid collecting device, a chevron
(vane) type or chimney tray type liquid collector (see Olujic
et al.1 or Rix and Olujic15) installed above a narrow trough gravity
liquid distributor.
For conventional packed columns and similar sections in a
packed DWC we have assumed a constant spacing of 1.8 m, which
is appropriate for the size of columns considered in this study. With
a symmetrical distribution of stages in two conventional columns,
each containing around 20 stages in rectication and stripping
sections, it was chosen to have two shorter beds per section. This
means that both packed columns contain three liquid redistribution
sections, with the middle one receiving also the feed.
Since the conventional columns and the DWC considered in
the present study operate at above atmospheric pressure and the
feed is a slightly subcooled liquid (q = 1.064), the critical loads
will be on the bottom stage of each column, which is therefore
taken as the basis for determining the diameter of the two conventional columns as well as the shell of a DWC. The characteristic values are shown in Table 3 together with the tangentto-tangent length (height) of the column shells for two columns
in conventional congurations, equipped with sieve trays and
B1-350MN packing, respectively.
4.1. Layout of a DWC Containing Structured Packings.
Arranging beds and liquid collecting and distribution become
more demanding when the partitioned part of a packed DWC is
considered. Figure 6 shows schematically the internal configuration of the DWC, indicating the number of stages in each column
section and the corresponding vapor and liquid loads at the top
and the bottom of each bed. Since the upper part of the DWC
contains 26 stages, this section consists of two beds, each
containing 13 stages. The bottom section contains 16 stages in
one bed. Due to a pronounced difference in vapor and particularly liquid loads below the feed (F) and side product (S) drawoff points, the partition wall in the lower part is in an off-center
position. That is, a much larger cross-sectional area is needed
below the feed point on the prefractionator side to accommodate
the feed stream that enters as slightly subcooled liquid. Adding

Table 3. Dimensions and Components of the DWC and Two Conventional Congurations
conguration
C1/C2 (trays)

C1/C2 (packings)

DWC (packings)a

column top pressure (bar)

1.7/2.7

1.7/2.7

2.7

shell diameter (m)

2/2

1.6/1.8

1.7

shell height (m)

40.5/39.5

27.1/27.5

37.3

number of trays or packed beds

61/59

4/4

7 (4)

number of distributors

4/4

7 (4)

number of liquid catchers


number of support grids




4/4
4/4

7 (4)
7 (4)

Numbers in parentheses indicate devices placed in partitioned part of the column.


5683

dx.doi.org/10.1021/ie1020206 |Ind. Eng. Chem. Res. 2011, 50, 56805692

Industrial & Engineering Chemistry Research

ARTICLE

Figure 7. Geometry factors associated with placing the partition wall.

Regarding the dimensioning procedure, the outgoing point


is the overall shell diameter determined using conventional
methods for the stage with the largest vapor load. The sum of
cross-sectional areas of the partitioned part of the column, which
however may dier above and below the feed and side product
draw-o stages, should be equal to the overall one, i.e.
Ac Apc Amc

where Ac (m2), Apc (m2), and Amc (m2) are the cross-sectional
areas of the shell of the DWC, prefractionator, and main column,
respectively. The cross-sectional area of the prefractionator
segment of the overall cross-sectional area can be calculated from
Figure 6. Internal conguration, indicating the number of stages in each
bed and the corresponding liquid and vapor ows at the top and bottom
of each bed.

the feed to the liquid coming from the bed above as well as a
certain amount of liquid obtained due to direct condensation of
the ascending vapor upon contact with subcooled feed liquid
makes the liquid load of the bed below the feed nearly 5 times
larger than that on the main column side. Also, on the main
column side of the partition wall there is a side draw where side
product is taken out of the column as a liquid. Owing to a rather
small vapor load this section requires a much lower crosssectional area than the heavily liquid loaded stripping section
on the prefractionator side. Although the vapor and liquid loads
in sections above the feed and draw-off differ to a certain extent,
the partition wall is, as indicated in Figure 6, placed in the center.
This appeared to be a convenient choice in the present case, and
placing a partition wall off-center, as required in the sections
below the feed and draw-off stages, can easily be arranged by
utilizing a nonwelded wall, which is a well-established practice in
the case of DWCs equipped with structured packings.16
One should note that welding a partition wall implies separating the shell into two semicircular sections with equal crosssectional areas. Owing to dierent loads in the prefreactionator
and the main column, this leads always to underdesign on the less
critical side. This was one of the main obstacles for wider
implementation of DWC in the beginning years, which, presently, can be circumvented by placing the partition wall ocenter, in the most appropriate way. Therefore, the introduction
of a nonwelded, self-xing partition wall in 1996 is considered to
be a milestone in the development of DWC technology.1

Apc

dc 2
 sin
8

where dc (m) is the dividing wall column shell diameter and


(deg) is the angle subtended by the partition wall (see Figure 7).
The latter is described as a function of the shell diameter and the
distance of the partition wall from the shell on prefractionator
side, lpcw (m):
2 arccos

dc  2lpcw
dc

The cross-sectional area of the main column is obtained


simply, by subtraction of prefractionator cross-sectional area
from the total one, using eq 1.
The obtained areas are then translated into diameters of an
equivalent cylindrical column. These are used in conjunction
with local vapor and liquid loads to check the upper limit with
respect to ooding condition. This should not exceed 80% of that
which would cause ooding. A good indication for this is the
loading point, i.e., the point of onset of loading, which can be
estimated with some condence using the appropriate equation
of the Delft model, shown later on. Practically, this means that,
the design point corresponds approximately with the ratio of
operating and loading point vapor loads, i.e., the corresponding
F-factors equal 1. However, the corresponding pressure drop
should not exceed 3 mbar/m. In other words, this pressure drop
is taken as the upper limit value for design purposes.
If the calculated pressure drop exceeds the limit in one of the
partitioned sections, a possibility is to use a coarser packing or the
same packing with a larger corrugation inclination angle. This is a
practical option if the bed height is lower than the section height,
5684

dx.doi.org/10.1021/ie1020206 |Ind. Eng. Chem. Res. 2011, 50, 56805692

Industrial & Engineering Chemistry Research

ARTICLE

Table 4. Basic Dimensions and Estimated Pressure Drop per


Packed Bed Section as Indicated in Figure 8

Section

Figure 8. Detailed drawing of the DWC with all major pieces of


equipment indicated.

which is often the case if a part of the partition wall is placed ocenter. Additional bed height is needed to accommodate the
required number of stages, because both a lower specic geometric area and an increased corrugation inclination angle are
associated with less eciency. Also the lateral position of the
partition wall can be corrected accordingly, provided the other
side can have it. If not, then the shell diameter of the DWC needs
to be enlarged accordingly. This means that a trial-and error
procedure is necessary to arrange proper lateral positioning of
the partition wall. The longitudinal positioning is related to the
number of stages contained on the prefractionator and the main
column side, above and below the feed and side product draw-o

2.1a 2.1b

(m)

5.2

(m2)

2.27 2.27

d
p

(m)
1.7 1.7
(mbar) 6.95 6.91

5.2

prefractionator

main column

1.1

2.2

1.2

2.3

2.4

3.6

5.2

4.8

4.0

1.135

1.599

1.135

0.671

6.4
2.27

1.20
5.75

1.43
7.95

1.20
4.25

0.92
9.76

1.7
13.51

stage, respectively. If one side contains more stages and the same
packing is chosen, then the side with fewer stages will have a
portion of empty space. This, however, allows installation of a
coarser packing, to arrive at a lower section pressure drop, if
appropriate. One should also note that additional height of the
column shell may be required to accommodate inclined portions
of the partition wall. This is necessary if the lateral positions of the
partition wall above and bellow the feed dier (one or both on
dierent o-center positions), as encountered in this study.
A reference design for the DWC was generated using the
Montz in-house method for establishing column dimensions.
The diameter of the DWC shell is 1.7 m, based on the design
point corresponding to 75% of the ooding vapor velocity on the
bottom stage.
A detailed drawing of this DWC is shown in Figure 8,
indicating seven packed beds and all auxiliary devices, i.e., column
internals of importance for cost estimation purposes, such as
liquid collectors or catchers, an externally placed liquid splitter,
the liquid distributors, the vapor distributor, and packing support
grids. In all situations a narrow trough liquid distributor with drip
tubes is used while the type of the liquid collector depends on the
specic liquid load. For the specic liquid loads above 20 m3/
m2 h, a chimney type collector is a preferred choice. This device is
also a common choice for side product draw-o location regardless of the liquid load. For lower specic liquid loads, vane
(chevron) type collectors are used. In present study a chimney
tray collector, placed above the vapor inlet, is used to facilitate
initial vapor distribution.
The proper distribution of vapor ow bellow the partition wall
is the key to success with the design of a DWC. This is something
that needs to be xed in the design phase, by arranging the
individual ow resistances to ensure equal pressure drop on two
sides of the partition wall. This is presently done by manufacturers only, i.e., using proprietary design methods. However, as
shown and demonstrated in what follows, it can be done with
required accuracy using recently published methods for estimating the pressure drop of corrugated sheet structured packings and
state-of-the-art packed column internals, respectively.
4.2. Pressure Drop of a Packed DWC. The heights of
individual column sections are indicated in Figure 8. The bed
heights and corresponding cross-sectional areas and equivalent
diameters for each section are shown in Table 4. The former
have been determined by multiplying the required number of
theoretical plates with a constant HETP value as adopted by
Montz for B1-350MN for this case (HETP = 0.4 m). One
should mention here that the Delft model arrives at conservative enough values, similar to those used by Montz in the
present case. However, a thorough verification effort is required
before adopting this method for the determination of bed
height of packed DWCs.
5685

dx.doi.org/10.1021/ie1020206 |Ind. Eng. Chem. Res. 2011, 50, 56805692

Industrial & Engineering Chemistry Research

ARTICLE

The heights and cross-sectional areas of individual beds allow


determination of the volume of packing and in conjunction with
average liquid and vapor loads serve as a basis for estimation of
the pressure drop, separately, for each bed and associated liquid
collecting and distributing equipment.
The estimated pressure drop for each of the beds contained in
two packed columns is shown in the bottom row of Table 4.
These numbers have been generated using the Delft model (see
Fair et al.17 or Olujic et al.18), which, as shown by Olujic et al.,19
can easily be adapted to account for any change in geometry of
corrugated sheet packings. Note that during the validation of this
model using the total reux data obtained with the chlorobenzene/ethylbenzene system,14 it appeared that the trend of
pressure drop curves is reproduced very well; however, the
predicted values appeared to be consistently on the lower side,
and the most pronounced deviation was observed at the highest
test pressure. Therefore, in the present study concerned with
above atmospheric operating pressures, the estimated values
were multiplied by the factor 2 to generate conservative predictions, resulting in numbers somewhat larger than those estimated
using the Montz in-house method.
4.2.1. Pressure Drop of Corrugated Sheet Structured Packings. In order to enable direct implementation of the proposed
DWC dimensioning method, the present paper contains necessary working equations of the Delft model, accounting, where
appropriate, for macro geometry modifications as implemented
in high performance structured packings, such as Montz-pak
B1-350MN considered in this study. This packing contains a
bend in the bottom part of corrugations and a corrugation angle
lower than the common 45.
A specic feature of the Delft model is that it makes a
distinction among three characteristic angles: the corrugation
inclination angle, R (deg), that strongly inuences the interaction
of gas streams at the crossing planes formed between ow
channels oriented in opposite directions; the eective gas ow
direction change angle at the transition between packing element
or layers, RDC (deg); and the eective ow angle of liquid, RL
(deg), which is steeper than the corrugation inclination angle due
to a more or less pronounced tendency of the liquid to ow over
the corrugation ridges driven by gravity. The latter two depend
on the corrugation inclination angle and can be determined using
expressions given later on, where appropriate. The corrugation
inclination angle is a characteristic geometric feature of corrugated sheet structured packings. The standard angle is 45.
Another commonly used (where appropriate) angle is 60, with
respect to the horizontal. Experimental evidence on the eect of
the corrugation inclination angle on the mass transfer eciency,
pressure drop, and capacity of corrugated sheet structured
packings can be found elsewhere, e.g., Olujic et al.19,20
4.2.1.1. Basic Geometry and Flow Related Parameters. In
addition to the corrugation inclination angle, R (deg), other
relevant macro geometric features of a corrugated sheet structured packing are the specific geometric area, ap (1/m), the void
fraction or porosity, (m3/m3), and the height of a packing
element, hpe (m). The requested bed height, hpb (m), is arranged
by placing the required number of packing elements or layers,
npe, above each other:
hpb npe hpe

Since the subsequent element/layers are usually rotated to each


other by 90, to maximize lateral spreading and mixing of both

phases, the number of packing elements or layers corresponds


with the number of flow direction changes the vapor phase makes
while ascending through the packed bed.
For a packing with a smooth bend in the bottom part of the
corrugations, the vapor ow direction change angle is simply
90 R
5
2
Assuming that only gravity and the corrugation shape aect
the liquid ow, the eective liquid ow angle, RL (deg), can be
described by
3
2
RDC

6
6
RL a tan6
4

7
cos90  R
7

 7
b 5
sin90  R cos a tan
2h

where h (m) represents the height and b (m) is the width of the
base of corrugations. The corresponding length of the corrugation side, s (m), follows from
r
b2
s
h2
7
4
These three basic corrugation geometry dimensions depend
on the specic geometric area of the packing, and, being highly
inuential, should be measured upon delivery of a packing, as
well as the corrugation inclination angle, to avoid uncertainties
later on. Assuming a standard conguration utilizing a crimp
angle of 90, which implies that b = 2h, the following expression
can be used to determine the installed specic geometric area or
to back calculate corrugation dimensions from the given area.
4s
8
bh
Another important geometry-related parameter is the
V-shaped fraction of the cross section of a triangular gas ow
channel occupied by liquid lm:
ap

2s
b 2s

A geometry-related parameter of general importance is the


hydraulic diameter of the triangular gas ow channels. Assuming
uniform liquid distribution, i.e., constant lm thickness, (m)

dhG

bh  2s2
bh
10
"
2 
 #0:5
bh  2s
bh  2s 2
bh  2s

2h
b
2h

Since the lm thickness encountered in practice and its variations


are rather small, this complex expression can be replaced without
introducing a signicant error by a much simpler one, valid for
dry channels.
dhGdry

2bh
b 2s

11

In addition to packing geometry related parameters, R, h, b,


hpe, npe, and packing porosity, , the Delft model requires also
information on mass ow rates, densities, and viscosities of two
phases, i.e., MG (kg/s), ML (kg/s), FG (kg/m3), FL (kg/m3), G
(Pa s), and L (Pa s), respectively. These operating parameters
5686

dx.doi.org/10.1021/ie1020206 |Ind. Eng. Chem. Res. 2011, 50, 56805692

Industrial & Engineering Chemistry Research

ARTICLE

need to be provided for the top and bottom of each bed contained, and this information can easily be retrieved from the
results of detailed performance calculations.
Corresponding supercial velocities are dened as
uGs

4MG
FG
2 p
FG
FG dc

12

and
uLs

4ML
FL dc 2

13

where dc (m) denotes the diameter of the column or the equivalent


diameter of a section in the partioned part of the column.
FG (Pa0.5), which is generally known as the F-factor, represents
the vapor load of the column section. Per denition
FG uGs FG

0:5

14

The F-factor is a highly practical design and operating parameter


(square root of the vapor ow kinetic energy), because it places all
distillation and similar column applications within a rather narrow
range of absolute values (roughly 0.54 Pa0.5) regardless of the
operating pressure.
Due to the inclination of the gas ow channel within a packing
element or layer, the length of the vapor ow path is larger than
the bed height, and the extent of this depends on the corrugation
inclination angle. This also means that eective velocities of
vapor and liquid are larger than the corresponding supercial
velocity, and additional enhancement depends on the reduction
in ow cross-sectional area due to packing porosity and the liquid
holdup, hL (m3/m3). The working expressions for respectively
eective vapor and liquid velocity account for these eects:
uGs
uGe
15
 hL sin R

Here, Fload is an empirical correction factor that describes the


amount of pressure drop enhancement within the loading region
with respect to the preloading region pressure drop.
!2=sin R
!0:13
FG
u2Ls
20
Fload 3:8
FG, lp
2 gdhG
This correction is activated when FG/FG,lp > 1, i.e., when the
operating vapor load exceeds that corresponding to the point of
the onset of loading, FG,lp, or, in other words, the point of departure from preloading conditions. The loading point F-factor is
described reasonably well by the following empirical expression
introduced by Verscho et al.,21 which accounts explicitly for a
pronounced eect of ow direction change angle.
0
@0:0532 gdhG

FL  FG
FG

FG, lp

10:57
r!0:25
FL
p
uLs
sin RDC 1:24 A
FG
FG

21
This critical load corresponds roughly with the design point,
which is usually set to 0.70.8 of the ood point. However, as
mentioned before, if the pressure drop corresponding to the
loading point condition FG/FG,lp = 1 exceeds 3 mbar/m, the
latter should be taken as the design limit.
According to the Delft model, the preloading region pressure
drop consists of three major contributions: the pressure drop
associated with the vaporliquid interaction at the interface,
pGL; the pressure drop associated with the vaporvapor
interaction at crossings of open sides of triangular gas ow
channels, pGG; and the pressure drop associated with the
change in ow direction at the transitions between subsequent
packing elements or layers, pDC.
ppreload pGL pGG pDC

and
uLe

uLs
hL sin RL

16

Assuming uniform wetting, the liquid holdup is simply dened


as a product of lm thickness, (m), and the specic geometric
area of the packing, ap (m2):
hL ap

17

This simple basic expression appeared to hold well in practice, in


conjunction with well-known Nusselt laminar falling lm thickness expression adapted for inclined walls:
!1=3
3L uLs
18

FL gap sin RL
where g (m2/s) is gravity acceleration and RL (deg) is the
eective angle of liquid lm ow, dened by eq 6.
4.2.1.2. Working Pressure Drop Model Equations. The Delft
model makes a distinction between the preloading region, where
film flow conditions prevail, and the loading region, where a more
complex fluid-dynamic situation is encountered.
 
p
p

Fload
19
z
z preload

GL GG DC

FG u2Ge
2

22

Three major sources of ow resistance are expressed in terms


of characteristic overall interaction coecients. The overall
gasliquid interaction coecient is given by
GL jGL

hpb
dhG sin R

23

The friction factor, GL, described by the well-known Colebrook and White expression,22 accounts for the eect of the
roughness of the interface, which is here assumed to be equal
to lm thickness.



2
=dhG 5:02
=dhG 14:5


log
GL 2 log
ReGrv
ReGrv
3:7
3:7
24
Here, ReGrv represents the relative phase velocity Reynolds
number:
ReGrv
5687

FG uGe uLe dhG


G

25

dx.doi.org/10.1021/ie1020206 |Ind. Eng. Chem. Res. 2011, 50, 56805692

Industrial & Engineering Chemistry Research

ARTICLE

The gasgas interaction coecient appeared to be a rather


strong function of the corrugation inclination angle.
GG 1  jGG

hpb
dhG sin R

1  j0:722cos R3:14

hpb
dhG sin R

26

The direction change loss coecient expression makes a


distinction between bulk and periphery of the packing. That is,
in the core of a packed bed the gas ow changes direction at
transitions between packing elements or layers only. However, in
the wall zone this happens also with channels ending at the wall,
which means that here ow direction changes also within the
height of a packing element. This appeared to be a reason for
increased pressure drop observed with the same packing in
columns with small diameters.23
DC

hpb
wall
hpe bulk

with
h2pe
2hpe
2

d

c
tan2 R
dc 2 tan R

!0:5

Table 5. Pressure Drop Caused by Narrow Trough Distributors (nt) Installed in Dierent Positions (1, 2, 4, 5, 8, 9, 12)
along the DWC, with Indication of Free Area and Corresponding Vapor Load

27

The pressure drop due to direction changes in the bulk


appeared to be a strong function of corrugation inclination angle
only:

0:445
4092u0:31
Ls 4715cos RDC
ReGe
0:779
34:19u0:44
Ls cos RDC

ReGe

12

type

nt

nt

nt

nt

nt

nt

nt

jnt

0.40

0.40

0.40

0.40

0.40

0.40

0.4

FG (Pa0.5)

1.16

1.23

1.41

1.14

1.00

1.65

1.40

p (mbar)

0.03

0.04

0.05

0.03

0.02

0.07

0.05

2
3

prefractionator

main column

10

11

13

uLe
type

m /m h 16.5 15.5
cc cc

8.9
cc

39.3
ct

15.8
ct

17.9
cc

34.7
ct

jcc/ct

0.25 0.25

0.19

0.25

0.25

0.25

0.25

FG

Pa0.5

1.27 1.20

1.25

1.34

0.93

1.63

1.57

mbar

0.36 0.32

0.68

0.49

0.24

0.60

0.54

nt 1:21:5  j2:5  j

33

chevron type liquid collector:


cc 1:52:5  2:5j

30

where ReGe represents the Reynolds number based on the


eective gas velocity.
F uGe dhG
G
G

narrow trough liquid distributor:

In the wall zone also the liquid and vapor load play an inuential
role:
wall

Position

28

29

main column

Table 6. Pressure Drop Caused by Chevron (cc) and Chimney Tray (ct) Collectors Installed in Dierent Positions (2, 3,
6, 7, 10, 11, 13) along the DWC, with Indication of Free Area
and Corresponding Vapor and Liquid Loads



hpe
2
arcsin

dc tan R

bulk 1:76cos RDC 1:63

prefractionator

chimney type liquid collector:


ct 1:21 2:51  j

31

The relative magnitudes of the three dierent pressure drop


components depend on the corrugation inclination and ow
direction change angles involved. A detailed numerical illustration
of exhibited eects can be found elsewhere, e.g., Olujic et al.19
4.2.2. Pressure Drop of Liquid Distributors and Collectors.
The pressure drop caused by liquid distributors and collectors
can easily be determined using the recently proposed model by
Rix and Olujic:15
!
cc, ct, nt FG2
32
pint
j2cc, ct, nt 2
where is the characteristic flow resistance coefficient and j is
the free (void) area, while indices cc, ct, and nt denote respectively the chevron type liquid collector, the chimney tray liquid
collector, and the narrow trough liquid distributor.
The characteristic ow resistance coecient is expressed as a
function of free area using the following expressions.

34

35

The free areas of all (narrow trough) liquid distributors and


corresponding vapor loads expressed as characteristic F-factors
are shown in Table 5, while the same can be found for dierent
types of liquid collectors in Table 6. The estimated pressure drop
for each device is shown in the bottom row of Tables 5 and 6,
respectively.
One should mention here that the above expressions account
for the observed liquid load eect and according to given
constant (representative) values of coecients the extent of this
eect is most pronounced in the case of the chevron collector.
The chosen value (1.5), however, may appear conservative in the
case of rather low liquid loads (below 3 m3/m2 h). The
experimental data that have served as the basis for development
and validation of the above expressions can be found elsewhere,
e.g., Rix and Olujic.15
The results are summarized in Table 7, indicating the relative
magnitudes of the pressure drops of internals and the partitioned
part of the column with respect to the total pressure drop. One
should note that in the present case a constant free area (25%) for
5688

dx.doi.org/10.1021/ie1020206 |Ind. Eng. Chem. Res. 2011, 50, 56805692

Industrial & Engineering Chemistry Research

ARTICLE

all liquid collectors was chosen, except for the chimney collector
on the main column side, which was reduced to 19% to ensure
equal pressure drop on both sides of the partition wall.
Being rather low, the pressure drop of support grids is
neglected. Also, the numbers shown in Tables 5 and 6 indicate
that the pressure drop of narrow trough liquid distributors is
much lower than that of liquid collectors. This is mainly due to a
much larger free area available for ascending vapor than in the
case of liquid collectors that require relatively more crosssectional area for liquid handling. In any case, as demonstrated
here, the free area of liquid collectors can serve as a means for
ne-tuning the pressure drop of sections in the partitioned part of
a DWC. This can be done automatically using Excel solver, which
will indicate the free area bound between 5 and 30%, generating
the requested pressure drop. The dimensioning procedure for a
three-product DWC is shown schematically in Figure 9.

5. ECONOMIC EVALUATION AND COST ESTIMATION


PROCEDURES
Total annualized costs (TACs) are taken as the basis for
evaluation of the economic feasibility of a DWC. Since dierent
Table 7. Overall Pressure Drop (p) of Packings and Column Internals as Well as That of a DWC with Indication of the
Contribution of Either of Two Sections in Partitioned Part of
the Column

congurations are compared on the same basis, TACs are


represented simply as a sum of the annual utility cost and 10%
of the installed equipment cost. The latter is based on the
assumption of a plant nancial lifetime of 10 years. The yearly
operation time is taken to be 8322 h, as encountered at INA. To
be closer to the common European and United States situation,
the following utility prices were considered in the present study:
US $0.03/t for cooling water, US $13/t for steam, and US $130/t
for fuel oil. The latter is used because a red heater is necessary in
conjunction with a rather high temperature of the bottoms of the
platformate splitter (rst column in the conventional sequence)
and the DWC.
The capital costs in the present cases include those associated
with column shell, trays, and/or packings and, for packed
columns, also liquid collectors, liquid distributors, and support
grids. Important external equipment components of each distillation column are the reboiler, condenser, and reux accumulator, respectively. The latter has not been considered in the
present study. In order to bring the installed equipment costs
estimated using the SI unit version of correlations from the
Douglas textbook24 to the price level corresponding to the year
2009, the corresponding annual value (1468.6) of the Marshall &
Swift Equipment Cost, published in the March 2010 issue of
Chemical Engineering,25 has been used.
The installed cost of a column shell (in US $) made of carbon
steel is estimated using the correlation:

p (mbar)

Cshell

prefractionator

main column

mbar
mbar

41.07
2.73

41.38
2.41

total

mbar

43.80

43.80

partitioned section

mbar

14.95

14.95

packings
internals



1468:6
fp
dc 1:066 h0:802
c, t-t
280

36

where dc (m) is the column diameter and hc,t-t (m) is the tangentto-tangent column height, while the cost factor fp depends on the
operating pressure. In the present case, for p e 3.5 bar, fp =
2981.68.

Figure 9. Schematic illustration of the DWC dimensioning procedure.


5689

dx.doi.org/10.1021/ie1020206 |Ind. Eng. Chem. Res. 2011, 50, 56805692

Industrial & Engineering Chemistry Research

ARTICLE

Table 8. Unit Prices of Structured Packing, Packed Column


Internals, and Sieve Trays
device

unit price

Table 9. Equipment Costs, Utility Costs, and Total Annualized Cost (TAC) for Conventional Sequence with Respectively Trayed and Packed Columns and a DWC (Price
Reference December 2009)

600 US $/m2

sieve trays
2

sructured packing (250 m /m )


liquid distributor

2000 US $/m3
4000 US $/m2

liquid collector

2000 US $/m2

support grid

800 US $/m2

For the purchased cost for sieve trays (carbon steel) as well as
structured packings and internals (both made of stainless steel),
we have used the values approved by Montz as reasonable
estimates for the purposes of this comparative study. These are
summarized in Table 8. For structured packings, the price is
proportional to the specic geometric area of the packing. This
means that in the present case the value given in Table 8 needs to
be multiplied by the factor 1.4, which is the ratio of specic
geometric area of packing used in this study, i.e., B1-350 MN, and
the basic type B1-250. That is, the given numbers represent the
nominal specic geometric area of a packing in m2/m3.
Given values are valid for conventional tray and packed
columns. In order to account properly for complexities associated with building a packed DWC, the cost of both packings
and related internals installed in the partitioned part is increased
by 20%, i.e., the value for standard equipment multiplied by the
factor 1.2.
The number of beds and their volumes and the number of
distributors, collectors, and support grids contained in conventional and partitioned parts of the column can be retrieved by
careful inspection of the detailed drawing shown in Figure 8.
Finally, to translate the purchased cost into installed cost for sieve
trays and for structured packings and related internals factors 3
and 2 are used, respectively.
For shell and tube condensers and reboilers, the following
expression allows estimation of the installed cost in US dollars:


1468:6
37
ctype A0:65
Ccond
280
where A (m2) is the required heat transfer area and ctype is the
coecient depending on the type of heat exchanger. For condensers, ctype = 1609.13, and for kettle reboilers, ctype = 1775.26.
These numbers are valid for common construction materials and
the operating pressure as encountered in the present case.
The installed cost (in US dollars) of a red heater is estimated
from


1468:6
14390:71Q 0:82 1:23 ft fm fp  38
Creb
280
where Q (MW) is the reboiler duty, ft is the correction factor
accounting for the type of heater (for a cylindrical heater, ft = 0),
fm is the correction factor that accounts for the construction
material (for carbon steel, fm = 0, and for stainless steel, fm = 0.5),
and fp is the correction factor for the eect of the operating
pressure (for pressures below 34 bar, fp = 0).
A summary of the capital, operating, and total costs for two
conventional congurations and a DWC is given in Table 9.
Although the conventional conguration with two columns
equipped with structured packings appears more cost-eective
than the conventional conguration employing tray columns, a

conguration
C1C2 (tray) C1C2 (packed) DWC (packed)
Installed Equipment Costs (US $)
column shell

1,261,781

781,468

501,621

column internals

678,240

611,793

516,332

reboiler

443,109

401,809

259,461

condenser

399,610

386,898

203,090

total capital

2,782,740

2,181,967

1,480,504

savings (%)

21.6

46.8

Operating Costs (US $/year)


cooling water
fuel oil/4 bar steam

121,834
938,071

119,337
907,096

59,169
452,491

total utilities

1,059,905

1,026,433

511,660

savings (%)

3.2

51.7

TAC (US $/year)


savings (%)

1,338,179

1,244,630

659,710

7.0

50.7

compact DWC is by far the most attractive option. Compared to


the conventional tray column conguration, a DWC would
require 46.8% less capital and 51.7% less utilities, which combined results in a 50.7% lower total annualized cost (TAC). This
is really appealing, indicating that implementation of a DWC
could be highly rewarding, i.e., could lead to a substantial increase
in protability of aromatics processing plants.

6. CONCLUDING REMARKS
In this paper we have demonstrated that a commercial
simulator can be used in conjunction with initial guesses for
governing variables obtained from a simple but theoretically
founded short-cut method to generate without computation
diculties an optimized internal conguration of a DWC.
Compared to the conventional two-column-in-series conguration for obtaining benzene- and toluene-rich fractions from a 15component feed, a DWC requires 43.3% less energy to get the
same product specications.
The Delft model for structured packings in combination with
the Rix and Olujic method for packed column internals proved to
be an eective and reliable tool for preliminary dimensioning of
DWCs equipped with both conventional and high performance
corrugated sheet structured packings. The free area of liquid
catchers appeared to be a suitable variable for tuning the pressure
drop in the partitioned part of a DWC.
The compact dimensions of a DWC translate into a considerably lower installed equipment cost. Expressed in total annualized costs (TACs), a DWC enables a 50.7% savings with respect
to the conventional two-columns-in-series conguration employing tray columns, and 47.0% savings with respect to those
employing the same type and size of structured packings.
The fact that much less energy, capital, and space is needed
makes a DWC a highly interesting option for implementation in
aromatics processing plants.
5690

dx.doi.org/10.1021/ie1020206 |Ind. Eng. Chem. Res. 2011, 50, 56805692

Industrial & Engineering Chemistry Research

AUTHOR INFORMATION
Corresponding Author

*E-mail: z.olujic@tudelft.nl.

ACKNOWLEDGMENT
The authors thank J. Montz for providing the reference
designs for conventional columns and for the corresponding
DWC as well as the information on purchased and installed costs
of trays, packed column internals, and structured packings. We
would also like to thank INA Renery Sisak, Croatia (the MOL
group), for providing actual plant data.
NOMENCLATURE
A = cross-sectional area (m2)
ap = specic geometric surface area of packing, m2/m3
ae = eective (interfacial) area, m2/m3
b = corrugation base length (m)
C = installed cost of equipment (US $)
ctype = heat exchanger type cost coecient
d = diameter (m)
dhG = hydraulic diameter for the gas phase (m)
FG = uGs(FG)0.5 = gas load factor (Pa0.5 or m/s (kg/m3)0.5)
FG,lp = loading point gas load factor (m/s (kg/m3)0.5)
Fload = loading eect factor
f = cost-related correction factor
g = gravity acceleration (m/s2)
HETP = height equivalent to a theoretical plate (m)
h = corrugation height (m)
hc,t-t = tangent-to-tangent column height (m)
hL = operating liquid holdup (m3 of liquid/m3 of bed)
hpb = height of the packed bed (m)
hpe = height of the packing element (m)
htray = height between top and bottom tray (m)
lG,pb = total length of gas ow channel in a packed bed (m)
lG,pe = length of gas ow channel in a packing element (m)
lpcw = distance from column shell to partition wall on prefractionator side (m)
MG = mass ow rate of gas/vapor (kg/s)
ML = mass ow rate of liquid (kg/s)
npe = number of packing elements (layers) in a bed
p = operating pressure (bar)
P = pressure drop (Pa or mbar)
Q = reboiler duty (MW)
ReGe = eective gas phase Reynolds number
ReGrv = relative velocity Reynolds number
ReL = Reynolds number for the liquid
s = corrugation side length (m)
TAC = total annualized cost (US $)
uGe = eective gas velocity (m/s)
uGs = supercial gas velocity (m/s)
uLe = eective liquid velocity (m/s)
uLs = supercial liquid velocity (m/s)
z = unit bed height (m)
Greek Symbols

R = corrugation inclination angle (deg)


RL = eective liquid ow angle (deg)
RDC = ow direction change angle (deg)
= liquid lm thickness (m)
= packing porosity (m3 of voids/m3 of bed)

ARTICLE

= column internals ow resistance coecient


DC = overall coecient for direction change losses
GG = overall coecient for gasgas friction losses
GL = overall coecient for gasliquid friction losses
= angle subtended by partition wall (deg)
G = viscosity of gas (Pa s)
L = viscosity of liquid (Pa s)
bulk = direction change factor for bulk zone
GG = gasgas friction factor
GL = gasliquid friction factor
wall = direction change factor for wall zone
FG = density of gas (kg/m3)
FL = density of liquid (kg/m3)
j = fraction of the triangular ow channel occupied by liquid
j = free or void area of column internals
= fraction of gas ow channels ending at column walls
Subscripts

cond = condenser
c = column
cc = chevron collector
ct = chimney tray collector
DC = direction change
GG = gasgas interaction
GL = gasliquid interaction
L = liquid
lam = laminar ow
m = related to construction material
mc = main column
nt = narrow trough distributor
o = overall
p = related to operating pressure
pc = prefractionator column
reb = reboiler
shell = related to column shell
t = related to red heater type
trays = related to trays
turb = turbulent ow

REFERENCES

 .; Jodecke, M.; Shilkin, A.; Schuch, G.; Kaibel, B.


(1) Olujic, Z
Equipment improvement trends in distillation. Chem. Eng. Process.
2009, 48, 10891104.
(2) Hartman, E. L.; Buttridge, I.; Wilson, D. F. Increase aromatics
complex protability. Hydrocarbon Process 2001, 80 (4), 97100.
 . Dividing wall column
(3) Dejanovic, I.; Matijasevic, Lj.; Olujic, Z
a breakthrough towards sustainable distilling. Chem. Eng. Process. 2010,
49, 559580.
(4) Triantafyllou, C.; Smith, R. The design and optimization of fully
thermally coupled distillation columns. Chem. Eng. Res. Des. 1992, 70,
118132.
(5) Segovia-Hernandez, J. G.; Hernandez, S.; Rico-Ramirez, V.;
Jimenez, A. A comparison of the feedback control behaviour between
thermally coupled and conventional distillation schemes. Comput. Chem.
Eng. 2004, 28, 811819.
(6) Kaibel, G.; Miller, C.; Stroezel, M.; von Watzdorf, R.; Jansen, H.
Industrieller Einsatz von Trennwandkolonnen und thermisch gekoppelten Destillationskolonnen. Chem.-Ing.-Tech. 2004, 76, 258263.
(7) Asprion, N.; Kaibel, G. Dividing wall columns: Fundamentals
and recent advances. Chem. Eng. Process. 2010, 49, 139146.
(8) Rangaiah, G. P.; Ooi, E. L.; Premkumar, R. A simplied
procedure for quick design of dividing-wall columns for industrial
applications. Chem. Prod. Process. Model. 2009, 4 (1), 142 (http//
www.bepress.com/cppm/vol4/iss1/7).

5691

dx.doi.org/10.1021/ie1020206 |Ind. Eng. Chem. Res. 2011, 50, 56805692

Industrial & Engineering Chemistry Research

ARTICLE

(9) Errico, M.; Tola, G.; Rong, B.-G.; Demurtas, D.; Turunen, I.
Energy saving and capital cost evaluation in distillation column
sequences with a divided wall column. Chem. Eng. Res. Des. 2009,
87, 16491657.
(10) Halvorsen, I. J. Minimum Energy Requirements in Complex
Distillation Arrangements. Thesis, Norwegian University of Science and
Technology (NTNU), Trondheim, Norway, 2001.
(11) Halvorsen, I. J.; Skogestad, S. Minimum energy consumption in
multicomponent distillation. 1. Vmin diagram for a two-product column.
Ind. Eng. Chem. Res. 2003, 42, 596604.
(12) Halvorsen, I. J.; Skogestad, S. Minimum energy consumption in
multicomponent distillation. 2. Three-product Petlyuk arrangements.
Ind. Eng. Chem. Res. 2003, 42, 605615.
 . Dividing wall column
(13) Dejanovic, I.; Matijasevic, Lj.; Olujic, Z
application to platformate splittera case study. 20th European Symposium on Computer Aided Process Engineering; Pierucci, S., Buzzi Ferraris,
G., Eds.; Elsevier: Amsterdam, 2010; pp 655660.
 .; Rietfort, T.; Jansen, H.; Kaibel, B.; Zich, E.; Ruert,
(14) Olujic, Z
G.; Zielke, T. Maximizing the performance of corrugated sheet structured packings. Proceedings of the 9th Distillation and Absorption
Conference, Sept 1215, 2010, Eindhoven, The Netherlands; 2010;
pp 605610.
 . Pressure drop of packed column internals.
(15) Rix, A.; Olujic, Z
Chem. Eng. Process. 2008, 47, 15201529.
 . Unxed
(16) Kaibel, B.; Jansen, H.; Rietfort, T.; Zich, E.; Olujic, Z
wall: the key to a breakthrough in dividing wall column technology.
Proceedings of Distillation Topical Conference, AIChE Spring National
Meeting, April, 2227, 2007, Houston, TX, USA; AIChE: New York,
2007; pp 2941.
(17) Fair, J. R.; Seibert, A. F.; Behrens, M.; Saraber, P. P.; Olujic, Z.
Structured Packing Performance: Experimental Evaluation of Two
Predictive Models. Ind. Eng. Chem. Res. 2000, 39, 17881796.
 .; Behrens, M.; Colli, L.; Paglianti, A. Predicting the
(18) Olujic, Z
eciency of corrugated sheet structured packings with large specic
surface area. Chem. Biochem. Eng. Q. 2004, 18, 8996.
 .; Behrens, M.; Spiegel, L. Experimental characterisa(19) Olujic, Z
tion and modelling the performance of a large specic area, high-capacity
structured packing. Ind. Eng. Chem. Res. 2007, 46, 883893.
 .; Seibert, A. F.; Fair, J. R. Inuence of corrugation
(20) Olujic, Z
geoemtry on the performance of structured packings: an experimental
study. Chem. Eng. Process. 2002, 39, 335342.
 .; Fair, J. R. A general correlation for
(21) Verschoof, H.-J.; Olujic, Z
predicting the loading point of corrugated sheet structured packing. Ind.
Eng. Chem. Res. 1999, 38, 36633669.
 . Compute friction factors fast for ow in pipes. Chem.
(22) Olujic, Z
Eng. 1981, 88 (25), 9193.
 . Eect of column diameter on pressure drop of a
(23) Olujic, Z
corrugated sheet structured packing. Chem. Eng. Res. Des. 1999, 77 (Part
A), 505510.
(24) Douglas, J. M. Conceptual Design of Chemical Processes;
McGraw-Hill: New York, 1988.
(25) Economic indicators. Chem. Eng. 2010, 117 (3), 64.

5692

dx.doi.org/10.1021/ie1020206 |Ind. Eng. Chem. Res. 2011, 50, 56805692

Das könnte Ihnen auch gefallen