Ind. Eng. Chem. Res.

1999, 38, 2973-2978

2973

PROCESS DESIGN AND CONTROL

A Novel Cross-Flow Cascade Packed Column Design and Simulation
Fengrong Chen,* Rongqi Zhou, Zhanting Duan, and Yumei Liu
Department of Chemical Engineering, Tsinghua University, Beijing 100084, Peoples Republic of China

This paper develops a novel cross-flow cascade packed column. The column has a smaller width
of packing in the direction of the gas flow and a bigger area for gas flow compared with a
conventional cross-flow device with identical column diameter, stage height, and cross-sectional
area for the irrigating liquid flows. The performance comparisons of the novel cross-flow cascade
packed column with a countercurrent packed column for stripping ethanol from water indicate
that the gas-phase pressure drop per theoretical stage in the novel cross-flow column is
significantly lower than that in a countercurrent packed column operation under identical
conditions. A methodology is developed for modeling the novel nine-stage cross-flow cascade
packed column by using the number of transfer unit technique. This method is demonstrated
by experimental data on ethanol-water.
Introduction
Gas-liquid contacting operations such as distillation,
absorption, and stripping are usually carried out in one
of three configurations, namely cocurrent, countercurrent, and crisscross-flow. In cocurrent operations, because of the small driving force for mass transfer, the
efficiency is generally small. In countercurrent operations the liquid phase runs down the packing by gravity
and the gas phase is forced upward. So the countercurrent device has more efficiency and high pressure drops.
In cross-flow operations the liquid downflow is vertical
while the gas phase is in cross-flow, alternating in a
back-and-forth direction. This operational mode offers
somewhat less efficiency and low pressure drops as
compared with a countercurrent process. However, it
is a more efficient mass-transfer device than a cocurrent
tower.
Countercurrent distillation is an effective unit process
widely used in petroleum refineries, organic synthesis,
fine chemicals, and so on. However, this process is not
superior when the efficiencies are based on pressure
drop, energy expended, or volume of packing, such as
the gentle separation of materials at the lowest possible
temperature. This work attempts to develop a new
cross-flow operation mode for improving pressure drop
and separation efficiency in the column based on the
traditional cross-flow operation.
A conventional cross-flow cascade column is shown
in Figure 1. The characteristics of cross-flow operations
have been discussed by many investigators.1-4 The key
factor that hinders the improvement of pressure drop
and separation effectiveness is the flow conditions of
liquid and vapor in the column, so improving the flow
modes of liquid and vapor in the column is the key to
* To whom correspondence should be addressed at Department of Chemical Engineering, University of Massachusetts,
Amherst MA 01003-3110. Phone: (413) 577-0137; fax: (413)
545-1647; e-mail: fchen@chevax.ecs.umass.edu.

Figure 1. A conventional cross-flow cascade packed column.

the solution to the problem. This paper describes a novel

cross-flow cascade packed column design. In particular,
this design makes it possible to improve the relation
between pressure drop and separation efficiency in a
vacuum distillation. In this research, an ethanol-water
mixture was used to check the separation behavior of
the novel cross-flow packed column. A methodology was
developed for predicting the stage efficiencies of separation in the novel cross-flow packed column by using the
number of transfer unit (NTU) technique and existing
correlations for mass-transfer coefficients in countercurrent packed columns.
Design of Device
A nine-stage cross-flow packed column is illustrated
in Figure 2. Table 1 contains the physical dimensions

10.1021/ie990109j CCC: $18.00 1999 American Chemical Society

Published on Web 07/03/1999

2974 Ind. Eng. Chem. Res., Vol. 38, No. 8, 1999

Table 1. Column Physical Dimensions
Na
9

h, m
0.1

A G , m2
3.927

10-3

AL, m2
2.945

10-4

AT, m2
4.909

10-4

w, m
7.5

10-3

ap, m2/m3
1.318

103

dp, m
5.0 10-3

a N, number of stages; h, height of stage; A , average of cross-sectional area for gas flow; A , cross-sectional area for liquid flow; A ,
G
L
T
total cross-sectional area of column; w, width of the packing in the direction of the gas flow; ap, total surface area of packing; dp, diameter
of sphere possessing the same surface area as a piece of rings packing; D0 ) 0.025 m; D1 ) 0.02 m; D2 ) 0.005 m.

Figure 2. Nine cross-flow staged packed column.

of the cross-flow packed column. In this column, liquid

goes vertically down from top to bottom through this
annular packing by cylindrical screens. Gas flows up
from the bottom of the column starting from the core of
the inner cylinder. Because of gas deflection baffles, gas
is forced to go through the packing to the outer cylinder
and then back again, alternating in a back-and-forth
direction, and finally goes out of the top of the column.
As illustrated in Figure 2, the gas flow mode is
different from that in a conventional cross-flow packed
column. The device designed in this study has smaller
width of packing in the direction of the gas flow and
bigger area for gas flow compared with a conventional
cross-flow device with identical column diameter, stage
height, and cross-sectional area for the irrigating liquid
flows. Thus the abundant area for gas flow will decrease
gas-phase pressure drop, and it can allow operation at
gas velocities above the flooding points of countercurrent
packed columns.
Experimental Section
The nine-stage cross-flow cascade column consists of
a hollow cylinder constructed of metal. The inner and
outer walls of the hollow cylinder are porous screens.
The ring packing with height equal to diameter filled
up the gap between the inner and outer screens in
random arrangement. The total packed height is 1 m.
The concentric plates and circular plates as gas deflection baffles positioned in the column on opposite sides
of the packed section divert the gas phase into the
packing, affecting a crisscross gas-liquid flow pattern.

Figure 3. Experimental setup for measurement of column

performance.

A schematic of the experiment setup for the measurement of column performance is given in Figure 3. The
flask was immersed in a stainless steel tank filed with
a 1:1 mixture of ethylene glycol and water as heating
fluid. The heating was controlled by a voltage controller.
The cooling water circulation for the condenser was
achieved by a circulator. The novel cross-flow column
and the countercurrent column with the same column
diameter (25 mm) and packed bed height (1 m) are used
to rate. The volume of packing for the novel cross-flow
column is 2.94 10-4 m3. The volume of packing for
the countercurrent column is 4.91 10-4 m3.
Table 2 shows the separation performance of the ninestage cross-flow packed column for ethanol-water
distillation. Table 3 shows the separation performance
of the countercurrent column for ethanol-water distillation. The pressure drop difference and the height
equivalent to a theoretical plate (HETP) difference for
these two columns are shown in Figure 4 and Figure 5.
Figures 4 and 5 show that the gas-phase pressure
drop of the conventional countercurrent packed column
is significantly higher than that of the cross-flow cascade
packed column operations. However, the efficiencies of
the novel cross-flow column were slightly lower than
that of the conventional countercurrent column operated
under identical conditions for the ethanol-water separation. The volume of packing for the novel cross-flow
column is only 60% of the countercurrent column.

Ind. Eng. Chem. Res., Vol. 38, No. 8, 1999 2975

Table 2. Separation Performance of Nine-Stage Cross-Flow Column for Ethanol-Water Distillation (Top Pressure ) 1
atm)
top composition of
bottom composition of pressure drop of
gas mass
expt. no. top temp., K bottom temp., K ethanol (molar fraction) ethanol (molar fraction) the column, Pa HETP, m velocity, kg/s
1
2
3
4
5

350.75
350.75
350.75
350.95
351.15

362.15
364.15
363.15
363.15
364.15

0.8218
0.8043
0.8049
0.7980
0.7862

0.0656
0.0467
0.0559
0.0588
0.0471

41.42
66.27
115.97
153.24
256.79

0.243
0.226
0.238
0.245
0.233

0.0942
0.1450
0.1883
0.2600
0.3833

Table 3. Separation Performance of Countercurrent Column for Ethanol-Water Distillation (Top Pressure ) 1 atm)
top composition of
bottom composition of pressure drop of
gas mass
expt. no. top temp., K bottom temp., K ethanol (molar fraction) ethanol (molar fraction) the column, Pa HETP, m velocity, kg/s
1
2
3
4
5

350.35
350.55
350.65
350.55
350.55

362.55
361.75
361.55
361.55
362.15

0.8907
0.8909
0.8825
0.8621
0.8907

0.0522
0.0629
0.0760
0.0736
0.0693

74.55
173.95
231.94
414.17
497.01

0.195
0.205
0.221
0.230
0.210

0.0875
0.1567
0.1808
0.2867
0.3250

Figure 4. Pressure drop difference for both columns.

Figure 6. Transfer unit in cross-flow and volume element h-w-z.

width w ) (D2 - D1)/2, and height h, shown in Figure

6. The width of the packed stage can be written as

w ) NOGHOG

(1)

Figure 5. HETP difference for both columns.

HOG ) G/(KOGaw)

(2)

Simulation

NOG )

yy

(3)

Model Equations. The simulation method of the

cross-flow packed column has been described by Thibodeaux et al.2,4 using the theoretical transfer units
concept. In this work, the NTU technique was applied
to the simulation of our nine-stage cross-flow cascade
packed column. The basic idea of the model is to define
the actual packed stage in the novel cross-flow packed
column as a transfer unit with length z ) (D1 + D2)/2,

dy
y* - y

out
A
in
A

where, HOG is the overall height of a transfer unit; NOG

is the number of overall gas-phase transfer units; G is
the superficial gas molar velocity; KOG is the overall gasphase mass-transfer coefficient; aw is the wetted interfacial area of packing; x is the liquid mole fraction; y is
the gas mole fraction; A is the component A in the
mixture.
The vapor-liquid equilibrium equation is given by

2976 Ind. Eng. Chem. Res., Vol. 38, No. 8, 1999

Rx
1 + (a - 1)x

y* ) f(x) )

(4)

To calculate NOG, we take the approximate equation4

to express the potential difference y* - y in eq 3 as
in
out
out
F(y) ) y* - y ) [f(xin
A ) - yA ]/2 + [f(xA ) - yA ]/2

(5)

Combining eq 3 with eq 5, we can obtain NOG as

NOG )

in
yout
A - yA

(6)

F(y)

We now proceed the calculation of HOG. Using twofilm theory5 shown in Figure 7, the mass flux of A can
be written as

NA ) KOG(yAG - y*), y* ) f(xAL)

(7)

NA ) KG(yAG - yAi)

(8)

NA ) KL(xAi - xAL)

(9)

Eqs 7 and 8 can yield

KOG )

KG(yAG - yAi)
yAG - y*

(10)

To calculate KOG we need to know yAi. According to

equilibrium eq 4, the calculation of xAi is necessary.
Combining eqs 8 and 9 with eq 4 yields

- KL(R - 1)x2Ai + [KGyAG(R - 1) - KGR - KL +

KLxAL(R - 1)]xAi + KLxAL + KGyAG ) 0 (11)
For calculating xAi of eq 11, one needs to have the
correlations for KG and KL. The correlations for KG and
KL obtained by Onda et al.6 were chosen:

( )( )
( )( ) ( )

KG ) 5.23

G
aPg

KL ) 0.0051

0.7

FgDgAB

2/3

L
awL

aPDgABP
(aPdP)-2
RT

1/3

FLDLAB

-1/2

FL
gL

(12)

-1/3

(aPdP)0.4
FL 1000
(13)
ML

aw ) aP 1 - exp -1.45ReL0.1Fr-0.05
We0.2
L
L

( ) ]}
L
c

(14)

FrL )

L
(Reynolds number)
aPL
aPL2
gF2L

(Froude number)

WeL )

L2
(Weber number)
aPLFL

(17)

As of now, the transport equations in both phases are

obtained. The steady-state model of the column is given
by the material and energy balances. For simplicity, we
assume that the molar heat of vaporization of the
mixture is constant, so the material and energy balances
can be decoupled. Thereby the material balance equations are only used to calculate the composition profiles
for the column. The material balance equations associated with any transfer unit, shown in Figure 6, are as
follows:
in
out
out
Lxin
A + GyA ) LxA + GyA

(18)

Algorithm. When the inlet liquid composition on the

top stage, the outgoing gas-phase composition from the
top stage, the superficial mass velocities for both phases,
the operating pressure in the column, the length, width,
and height of packing for each transfer unit shown in
Table 1 and Figure 6, and the existing correlations for
mass-transfer coefficients in countercurrent packed
columns are known, it is possible to calculate the column
profiles by using the algorithm described in the flow
sheet diagram given in Figure 8. The average value of
the stage compositions is used to calculate the thermal
and transport properties in both phases, as shown in
eq 19.
in
out
ave
in
out
xave
A ) (xA + xA )/2, yA ) (yA + yA )/2

(19)

-0.75

where

ReL )

Figure 7. Two-film theory.

(15)

(16)

Calculation Results and Discussion. Our experimental data on ethanol-water were used as the basis
of calculation for the novel cross-flow column. The inlet
liquid composition on the top stage and the outgoing
gas-phase composition from the top stage were obtained
from the experimental data, and the compositions at
each succeeding stage were calculated by the algorithm
in Figure 8. Table 4 shows the comparison of the
experimental data with the calculated values. The
calculated results are within 30% error of the experimental data. Because of ignoring the heat effects and
the wall effects in the novel column, the obvious error
of the simplified model was generated. Even though the

Ind. Eng. Chem. Res., Vol. 38, No. 8, 1999 2977

Table 4. Comparison of Experimental Results with Calculated Results
expt. no.
1
2
3
4
5
a

aL,

kg/(m3s)

0.3198
0.4923
0.6393
0.8828
1.3014

aG,

kg/(m3s)

0.0240
0.0369
0.0479
0.0662
0.0976

xBottom
ethanol(expt.)

xBottom
ethanol(calc.)

expt. - calc.

(expt. - calc.)/expt.

0.0656
0.0467
0.0559
0.0588
0.0471

0.0480
0.0437
0.0461
0.0463
0.0458

0.0176
0.003
0.0098
0.0125
0.0013

26.83%
6.42%
17.53%
21.26%
2.76%

L, superficial liquid mass velocity; G, superficial gas mass velocity.

Figure 9. Murphree efficiency profile for different superficial gas

mass velocity.

Murphree efficiency decreases dramatically from the top

to the bottom in the cross-flow column for a superficial
gas mass velocity. This implies a much lower HETP
toward the top of the column. This predicted result is
not surprising because the HOG data calculated at the
top of the column gives smaller values than are calculated at the bottom of the column. Because the liquid
surface tension of water is almost three times that of
ethanol at the same temperature and the concentration
of ethanol increases from the bottom to the top in the
column, the wetted surface area of the packing at the
top section is bigger than that at the bottom section,
from eq 14. Then we can obtain the smaller HOG values
at the top section on the basis of eq 2.
Conclusions and Future Work
Figure 8. Flow sheet of the simulation algorithm.

calculated results obtained from this simplified model

may not be qualitatively correct, they are expected to
preserve all the important qualitative features of this
operation.
To check the changes of the efficiencies of each
transfer unit for the different superficial gas mass
velocities, the Murphree efficiency for a single transfer
unit was used for calculations, as defined in eq 20.

E(Murp.) )

in
yout
A - yA
in
y*(xout
A ) - yA

(20)

On the basis of the experimental data in Table 2, the

Murphree efficiency profiles for different superficial gas
mass velocities in the novel nine-stage cross-flow cascade packed column for stripping ethanol from water
are shown in Figure 9. It was observed that the
Murphree efficiency decreases with an increase in the
superficial gas mass velocity for a single stage, and the

This work illustrates a novel cross-flow cascade

packed column operation. The novel cross-flow packed
column has smaller width of packing in the direction of
the gas flow and a bigger area for gas flow compared
with a conventional cross-flow device with identical
column diameter, stage height, and cross-sectional area
for the irrigating liquid flows. The experimental comparison of the novel cross-flow cascade packed column
with a countercurrent packed column for stripping
ethanol from water indicates that the gas-phase pressure drop in the novel cross-flow column is significantly
lower than that in a countercurrent packed column
under identical conditions. However, the efficiencies of
the novel cross-flow column were slightly lower than
those of the conventional countercurrent column. The
volume of packing for the novel cross-flow column is
significantly less than that for the countercurrent
column. So it is more constructive to consider this device
as a separate gas-liquid contactor and view it as a
possible alternative to tray devices and conventional
packed devices.

2978 Ind. Eng. Chem. Res., Vol. 38, No. 8, 1999

The simplified simulation algorithm has been proposed for the novel cross-flow cascade packed column
operation, and calculations based on this algorithm are
supported by the experimental results.
With these and other approaches, there is significant
opportunity for research in scale-up effects and developing new annular structured packing instead of current
annular random packing by cylindrical screens for the
novel cross-flow operation. A further rigorous model that
takes into account the heat effects and the wall effects
should be developed.
Nomenclature
aP ) total surface area of packing (m2/m3)
aw ) wetted surface area of packing (m2/m3)
D ) diffusivity (m2/s)
dP ) diameter of sphere possessing the same surface area
as a piece of rings packing
G ) superficial molar velocity of gas (mol/m2s)
G ) superficial mass velocity of gas (kg/m2s)
g ) gravitational constant, 9.81 (m/s2)
KOG ) overall gas-phase mass-transfer coefficient (mol/m2
s)
KG ) gas-phase mass-transfer coefficient (mol/m2s)
KL ) liquid-phase mass-transfer coefficient (mol/m2s)
L ) superficial molar velocity of liquid (mol/m2s)
L ) superficial mass velocity of liquid (kg/m2s)
ML ) molecular mass of liquid mixture (g/mol)
P ) pressure (N/m2)
R ) gas constant, 8.3145 (J/molK)
T ) temperature (K)
w ) width of the packing in the direction of the gas flow
(m)
x ) liquid mole fraction
y ) vapor mole fraction
y* ) mole fraction of vapor in equilibrium with liquid
composition
Greek Letters
R ) relative volatility
) tolerance
) viscosity (kg/ms)
F ) density (kg/m3)
c ) critical surface tension of packing material, 0.075 (N/
m)

) surface tension (N/m)

Subscripts
A ) component A
AB ) a mixture of A and B
G or g ) gas phase
i ) interface
L ) liquid phase
Superscripts
ave ) average
cal ) calculated
eq ) equilibrium
g ) gas phase
in ) liquid or gas in
out ) liquid or gas out
L ) liquid phase

Literature Cited
(1) Hayashi, Y.; Kawanishi, T.; Uda, T.; Akiyama, Y.; Kageyama, N. Performance of cross-flow cascade packed column at gas
velocities above the flooding point of countercurrent beds. Ind. Eng.
Chem. Res. 1993, 32, 652.
(2) Thibodeaux, L. J.; Daner, D. R.; Kimura, A.; Millican, J.
D.; Parikh, R. J. Mass transfer units in single and multiple stage
packed bed, cross-flow devices. Ind. Eng. Chem., Process Des. Dev.
1977, 16, 325.
(3) Thibodeaux, L J. Fluid dynamic observations on a packed,
cross-flow cascade at high loadings. Ind. Eng. Chem., Process Des.
Dev. 1980, 19, 33.
(4) Velaga, A.; Thibodeaux, L. J.; Valsaraj, K. T.; Eldridge, R.
B.; Moncada, D. M.; Cho, J. S. Packed crisscross-flow cascade tower
efficiencies for methanol-water separations. Experimental versus
calculated values based on countercurrent flow correlations. Ind.
Eng. Chem. Res. 1988, 7, 1481.
(5) Treybal, R. E. Mass-Transfer Operations, 3rd ed.; McGrawHill: New York, 1980.
(6) Onda, K.; Takeuchi, H.; Okumoto Y. Mass transfer coefficients between gas and liquid phases in packed columns. J.
Chem. Eng. Jpn. 1968, 1, 56.

Received for review February 12, 1999

Revised manuscript received May 19, 1999
Accepted May 25, 1999
IE990109J