You are on page 1of 9

1632

Ind. Eng. Chem. Res. 2002, 41, 1632-1640

Distillation Sieve Trays without Downcomers: Prediction of


Performance Characteristics
J. Antonio Garcia and James R. Fair*
Department of Chemical Engineering, The University of Texas at Austin, Austin, Texas 78712

A large amount of performance data on larger-scale trays without downcomers has become
available recently, and the data have been examined with a view toward understanding and
modeling the contacting mechanisms under distillation conditions. The objective of this paper
is to describe new predictive models for efficiency, pressure drop, and flooding of trays without
downcomers and to show how they can be applied to distillation separations. Such an effort has
not been reported previously. The contacting devices are assigned the generic name of dualflow trays. The models are shown to give reasonable estimates of the performance of these trays
in larger columns with open areas in the range of 15-25%, hole diameters in the range of 1225 mm, and tray spacings in the range of 0.3-0.6 m.
Bubble trays for distillation columns normally involve
liquid flowing between trays through connecting downcomers. Such contacting devices are of the cross-flow
type, and vapor flows only through dispersers on the
trays such as holes, valves, or bubble caps. Less wellknown and -specified are perforated trays without
downcomers, wherein the liquid and vapor flow countercurrently through the same tray openings. These are
often called dual-flow trays, but in specialty forms, they
have other names such as turbogrid trays and ripple
trays. The devices are used for special services, especially when openings of a cross-flow tray might foul.
Their less general use appears to have derived from an
expected narrow operating range of high efficiency, as
well as a general unavailability of design models that
can enable reliable prediction of their performance.
In the mid- to late-1950s Fractionation Research, Inc.
(FRI), undertook an extensive research program on
dual-flow trays, obtaining performance data for several
different systems but not providing a general, fundamental correlation of the data for pressure drop, efficiency, and flooding. Meanwhile, Shell completed an
extensive program of testing its turbogrid trays, similar
in characteristics to dual-flow trays. Recently, FRI
released all of its dual-flow test results, and these data
form a major basis for the modeling in the present
paper. Shell has released none of its turbogrid test
results.
Dual-flow trays have been applied to many situations
in which a broad operating range (high turndown ratio)
is not essential. In its range of application, the tray
provides a very high mass transfer efficiency with low
capital investment. Importantly, the application of such
devices to fouling systems has been eminently successful, the alternating vapor-liquid passage through the
holes providing a self-cleaning action.
Previous Work

The published reports cover tests in distillation columns


with diameters as large as 2.5 m (8.2 ft). Test data for
a 1.0-m (3.3-ft) column containing turbogrid trays have
been presented by researchers at the Institute of Process
Fundamentals in Prague.3-6 The FRI work was conducted in a 1.2-m (4.0-ft) research column, using several
hydrocarbon systems plus water/alcohol and water/
steam. Both Shell and FRI confirmed that the slotted
openings of the turbogrids had equivalent performance
characteristics to the round holes of dual-flow trays if
slots with equivalent diameters were used. The results
of the FRI tests have been released to the public and
are now available from Oklahoma State University.7
Very little has been done to model the mass transfer
performance of dual-flow trays. In the 1950s and 1960s,
work in Europe with very small columns led to attempts
at dealing with the hydraulics of the trays, but serious
scale-up problems were anticipated, and no further work
transpired that would aid the commercial-scale designer. The most recently reported attempt to model
dual-flow tray efficiency, by Xu et al.,8 appeared in 1994.
They studied methanol/water and methanol/2-propanol
distillations in a 0.3-m (1.0- ft) column and provided
valuable insights into the mass transfer relationships
involved. However, they made no attempt to combine
their data with those of others to arrive at a more
generalized treatment of dual-flow tray efficiency.
Perhaps a key finding, although nonquantitative, has
been that the insertion of dual-flow trays as replacements for cross-flow trays dramatically alleviates problems with severe fouling. According to Baird,9 his
company has successfully supplied our DualFlo trays
to over 70 installations where solids or scaling represented a potential operating problem. These trays,
which range from 6 in. to 7 ft (0.15 to 2.1 m) in diameter,
stay cleaner longer than other designs and are easier
to remove and clean. Such comments have encouraged
designers to use the devices, termed self-cleaning
trays, without adequate predictive model support.

Very few of the results of Shells investigations of


turbogrid trays have appeared in the open literature.1,2

Modeling Approach

* To whom correspondence should be addressed. E-mail:


fair@che.utexas.edu.
Current address: DuPont Engineering Technology, Brandywine 8226, 1007 Market Street, Wilmington, DE 19898.

Figure 1 shows a schematic of a dual-flow tray. Vapor


and liquid flow countercurrently, using the same openings, and a volume of aerated liquid called the froth zone
is generated above the tray perforations. It is in this

10.1021/ie010326w CCC: $22.00 2002 American Chemical Society


Published on Web 02/08/2002

Ind. Eng. Chem. Res., Vol. 41, No. 6, 2002 1633

Figure 1. Schematic of a dual-flow tray showing mass transfer


zones.

Figure 3. Efficiency of stamped turbogrid trays, benzene/toluene,


atmospheric pressure. Open area ) 14%, tray spacing ) 500 mm,
hole diameter ) 12 mm, total reflux. Data from ref 2.

Figure 2. Efficiency of stamped turbogrid trays, methanol/water,


1.0-m column, atmospheric pressure, total reflux. Data from ref
5.

zone that the majority of the mass transfer is presumed


to occur. The space above the froth, called the spray
zone, is available for additional mass transfer, and this
zone differs in performance from the equivalent zone of
a cross-flow tray.
Observations of operating dual-flow trays reveal a
dynamic contacting process in which a given opening
alternately passes vapor and liquid. This accounts for
the self-cleaning character of the device. At any instant,
a certain fraction of the holes are actively passing vapor,
either as jets or as bubbles. Efficiency profiles (Figure
2) show a strong modification of the hole dynamics to
the extent that there is less self-cleaning potential at
lower loadings and clearly less froth volume to accommodate mass transfer needs. At very high loadings,
liquid entrainment occurs, thereby reducing the mass
transfer efficiency. Figures 2 and 3 (refs 2 and 5) show
the sharp efficiency profiles characteristic of dual-flow
trays. An effect of column diameter is indicated in
Figure 3. Figure 4 shows profiles of both efficiency and
pressure drop for representative FRI tests.3 For all of
the figures, it is apparent that a peak efficiency exists,
and in the present paper, an attempt is made to model
this efficiency. Depending on the propensity for liquid
entrainment upward, the peak appears to be in the

Figure 4. Efficiency and pressure drop of dual-flow trays,


cyclohexane/n-heptane, tray spacing ) 0.61 m, hole diameter )
12.7 mm, column diameter ) 1.2 m, pressure ) 1.63 atm, total
reflux. FRI data.

range of 75-90% of the maximum throughput, i.e.,


where the efficiency becomes very low and the pressure
drop becomes very high. Another important objective
of the present effort is to develop a generalized relationship for predicting this maximum (flooding) condition.
Test conditions in the database (Table 1) include the
conditions for peak efficiency. The database covers a
broad range of physical properties.
Flooding. The FRI test data (see Table 1) include
measurements of the near-flood condition. As defined
by FRI, the flood point is evidenced by a significant
increase in liquid holdup (pressure drop). The true
flood with essentially no separation is at a loading
slightly higher than the reported flood point. The flood
data have been correlated in the same semiempirical
manner used successfully for cross-flow type trays,10 as
shown in Figure 5. The plotting method provides a very
good fit of the data, and this leads to some insights
regarding the mechanics of contacting. The capacity
factor, Csb, has been normalized to a tray spacing of 0.61

1634

Ind. Eng. Chem. Res., Vol. 41, No. 6, 2002

Table 1. Database, Tests on Dual-Flow Trays

refa

system

FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
5
FRI
FRI
FRI
2
FRI

C6/C7
C6/C7
C6/C7
C6/C7
C6/C7
C6/C7
C6/C7
C6/C7
C6/C7
C6/C7
C4
C4
C4
C4
C4
C4
C4
C4
C4
C4
IPA/H2O
xylenes
xylenes
C8/C10
C8/C10
MeOH/H2O
MeOH/H2O
MeOH/H2O
MeOH/H2O
Bz/Tol
Bz/Tol

pressure
(kPa)

TS
(m)

%
open

hole
diam
(mm)

28
28
28
166
166
166
166
166
345
345
1138
1138
1138
1138
1138
2073
2073
2756
2756
3456
101
16 mm
16 mm
10 mm
10 mm
101
101
101
101
101
101

0.61
0.61
0.61
0.61
0.30
0.91
0.61
0.61
0.61
0.91
0.61
0.30
0.61
0.61
0.91
0.61
0.61
0.61
0.61
0.61
0.61
0.61
0.61
0.61
0.61
0.41
0.41
0.41
0.41
0.51
0.51

13.5
18.9
25.6
13.5
19.1
19.1
18.9
25.6
18.9
29.3
13.5
19.1
18.9
25.6
29.3
13.5
29
13.5
29
13.5
19.1
12.9
17.8
12.9
17.8
10.5
14.2
18.2
23.6
14
14

12.7
25.4
25.4
12.7
12.7
12.7
25.4
25.4
25.4
11.9
0.5
1.0
1.0
1.0
0.47
0.50
0.47
0.50
0.47
0.50
0.5
12.7
12.7
12.7
12.7
b
b
b
b
12.7
12.7

Ep/Fs

Fs
flood
[m/s (kg/m3)0.5]

peak
efficiency
(% flood)

79/1.98
75/2.26
63/2.56
83/1.77
55/1.89
57/3.11
81/2.20
55/2.60
98/2.01
67/2.81
121/0.83
90/1.29
115/1.13
92/1.62
96/1.89
114/0.67
91/0.91
150/0.43
87/0.79
>180/0.24
75/2.34
81/2.01
63/2.01
72/2.81
42/2.56
95/1.27
82/1.92
82/1.98
53/2.3
87/1.63
80/1.96

2.20
2.59
2.93
2.01
1.98
3.39
2.34
2.89
2.34
3.17
1.43
1.46
1.77
1.73
2.07
0.76
1.01
0.52
0.88
0.33
2.46
3.07
3.3
(3.78)
3.2
1.87
2.66
2.78
NA
2.44
2.62

90
87
88
88
96
92
94
90
86
88
58
88
64
94
91
89
90
81
90
74
95
65
61
(74)
(81)
68
72
71
NA
67
75

Column diameters: FRI, 1.2 m; ref 5, 1.0 m; ref 2, 0.45 and 2.50 m. b Slots, 4 mm 150 mm.

Figure 5. Flooding capacity of dual-flow trays based on FRI data,


1.2-m column diameter.

m (24 in.). Thus, the 0.30-m spacing has about 81%, and
the 0.91-m spacing about 116%, of the capacity of the
0.61-m spacing.
The effect of open area on flooding is evident but not
well understood. Apparently, the higher hole velocity
for lower open area results in jetting, which increases
liquid entrainment. A small effect of hole size on
flooding is indicated, with 25-mm holes having about
6% less capacity than the more standard 12.5-mm holes.
This difference might be more apparent than real, but
it could be a result of the longer jet length for the larger
holes.
Efficiency at Lower Loadings. The profiles in
Figures 2-4 are typical of dual-flow trays, and it is clear

that, at lower loadings, there is a distinct loss of


efficiency, much like the case for cross-flow sieve trays
operating under conditions of weeping and dumping or
the case for packed columns with poor liquid distributions. Observations indicate that, in this region, a
definite loss of froth height occurs, fewer holes have
intermittent vapor and liquid flow, and the residence
time of vapor contacting liquid is shorter. The loss of
efficiency for weeping/dumping of cross-flow trays has
been interpreted in an approximate way by adapting
the Colburn11 relationship for liquid entrainment. The
same approach will be used for dual-flow trays, as
discussed later.
Liquid Entrainment in Vapor. The experimental
curves of Figure 2 are representative of the shapes of
the efficiency-velocity relationships found for all tests
of dual-flow trays. To use these curves as an example,
the efficiency at flow rates higher than the peak value
are subject to the same type of entrainment analysis
as used for cross-flow trays.10,12 The peaks occur at
about 1.10 and 1.85 m/s superficial velocity for the
10.5% open and 18.2% open trays, respectively. At
higher loadings, the effect of entrainment on efficiency
can be approximated by a modification10 of the Colburn
relationship

Ew/Ep )

1
1 + Ep/(1 - )

(1)

1-
Ep + (1 - )

(2)

or

where ) Ew/Ep, Ew is the wet efficiency, Ep is the


dry efficiency at the peak, and is the ratio of the

Ind. Eng. Chem. Res., Vol. 41, No. 6, 2002 1635

Figure 7. Effect of hole diameter on tray efficiency and pressure


drop of dual-flow trays. Cyclohexane/n-heptane at 1.63 atm and
total reflux. Tray spacing ) 0.61 m, open area ) 19.1%. FRI data.

drop curve, indicating the onset of loading where higher


vapor velocities cause increased liquid holdup. This is
analogous to loading in packed columns.
For a pressure balance to be maintained on a given
dual-flow tray

ht ) hgd + hL ) hL - hLd
Figure 6. Fractional entrainment of liquid in vapor. Methanol/
water, 1.0-m-diameter column. Data from ref 5 and Figure 2. The
superimposed curve for cross-flow sieve trays is taken from ref
12.

moles of liquid entrained to the total downflow including


entrainment recycle.
Representative values of calculated from the experimental data of Figure 2 are shown in Figure 6. Also
shown is a composite curve for cross-flow sieve tray
entrainment, taken from ref 12. It appears that similar
mechanisms for entrainment prevail for the two tray
types. Tentatively, the generalized curves for sieve trays
can be used for estimating the liquid entrained by the
rising vapor of dual-flow trays. This assumes that the
vapor breaks through a dual-flow tray froth in the same
manner that it breaks through a cross-flow sieve tray
froth. A more quantitative assessment of the entrainment effect must await experimental data; however,
prudent designs call for an approach to flooding not
greater than 80-85%, close to the peak efficiency region,
where the entrainment effect is minor.
Low Loading Effects. To interpret the loss of
efficiency at lower loadings, an entrainment-in-reverse
approach is used, even though there is no recycle due
to entrainment. To take an example from Figure 2, for
a superficial velocity of 1.10 m/s (1.10/1.85 or about 60%
of the peak loading), the efficiency is 58%, or 70% of
the peak value. Using an adaptation of eq 2, one obtains

1 -
1 - 0.70
)
) 0.34 (3)
Ep + (1 - ) 0.70(1 - 0.70)

where is the fraction of the peak efficiency (E/Ep)


and is a correlating parameter. Appropriate values
of thus enable the total efficiency curve to be
established.
Pressure Drop. Representative pressure drop data
for dual-flow trays of two open areas are shown in
Figure 4, and Figure 7 shows the effect of hole size on
pressure drop (as well as on efficiency). In the region of
peak efficiency, there is an evident break in the pressure

(4)

where ht is the total pressure drop across the tray,


height of clear liquid; hgd is the pressure drop for vapor
passing through fraction of holes x; hL is the residual
pressure loss for vapor flowing through the froth; hLd is
the pressure loss for liquid flowing through 1 - x
fraction of holes; and hL is the liquid head to force liquid
through 1 - x fraction of holes.
Equation 4 implies that hL is greater than hL, because
liquid and vapor flow in opposite directions through the
holes. The dynamics of the dual-flow tray are such that
momentum in the liquid is converted to static head to
allow liquid to flow downward against the static pressure gradient. Because the mechanisms of flow are so
poorly understood, we will concentrate on the gas flow
portion of eq 4

ht ) hdG + hL

(5)

with

hdG )

Ux2FL
2gCv2FG

(6)

where Ux is the linear velocity of vapor through x


fraction of total holes and Cv is the orifice coefficient, a
function of the hole diameter and effective open area.
Orifice Coefficient. For flow through holes in a dualflow tray, the correlation developed originally for sieve
trays by Leibson et al.,13 presented originally in graphical form, can be represented analytically by

Cv ) 0.74(Ah/Aa) + exp[0.29(tt/dh) - 0.56]

(7)

where tt is the tray thickness, dh is the hole diameter,


and the hole area Ah is based on the holes passing vapor.
Orifice coefficients obtained for the tray geometries
considered in this work (Table 1) ranged from 0.58 to
0.72. By comparison, for a sieve tray with 12% hole area,
12-mm holes, and a tray thickness of 1.6 mm, the orifice
coefficient has a value of 0.80. For the two hole sizes of

1636

Ind. Eng. Chem. Res., Vol. 41, No. 6, 2002

Figure 7, the difference in orifice coefficient accounts


for about one-third of the difference in total pressure
drop.
Liquid Holdup. Earlier work by Kotschering et al.14
provided a correlation for equivalent clear liquid height
on dual-flow trays, which was modified later by Xu et
al.8 The Xu modification was adapted to the FRI data
in the present work as follows

hL ) b1

(LML)n[US(FG/FL)0.5]b2
FL(Ah/Aa)b3 (tt/dh)0.42

(8)

with regressed values of the constants b1 ) 0.01728, b2


) 1.0, and b3 ) 1.50. In the original work, the exponent
n is obtained from the expression

n ) 4.3

()
Ah
Aa

1.5

(9)

Figure 8. Fraction of holes passing vapor, cyclohexane/n-heptane,


1.63 atm. Tray spacings, open areas, and hole diameters as
indicated. FRI data.

Fraction of Holes Passing Vapor. Because there are


no measurements of this parameter, it must be deduced
from measured pressure drop data for the entire tray
(eq 5). After correcting for liquid holdup hL by eq 8, eq
6 is used

hdG ) ht,meas - hL )

Ux2FL
2gCv2FG

(10)

from which

Q
Ah

FL

(ht,meas - hL)(2gCv2FG)

(11)

where Q is the volumetric flow rate of vapor.


Even though this approach is semiempirical, it is
useful for predicting tray pressure drops from the
number of active holes. The fraction has been correlated as a function of hole area, tray spacing, and
approach to flood using the FRI data as

( )( )

)A

Ah/At
0.2

0.8

TS
0.610

0.2

exp -0.35

(|% floodC - B|)]

Figure 9. Parity plot for the experimental and calculated


pressure drops of dual-flow trays. Cyclohexane/n-heptane, ibutane/n-butane, and 2-propanol/water (136 points).

(12)
where TS is the tray spacing in meters and A, B, and C
are the regression correlation constants (A ) 0.4668, B
) 90, C ) 45).
A representative plot of fractional holes passing vapor
is shown as Figure 8. Although the trend with loading
might seem counterintuitive, until direct measurements
are made, the relationships will remain useful for the
design process, as shown later in a comparison of
calculated and measured overall pressure drop data.
Pressure Drop Model Analysis. The approach
outlined above was used to develop the parity plot in
Figure 9. The plot is based on all of the total reflux FRI
experimental data. The mean absolute deviation is
23.0%, and the average deviation is 0%. Of the 136
points, 13 lie above the (25% range. Most of these
outside points derive from near-flooding conditions,
where measurements are not precise. If the water/2propanol system is excluded, leaving only hydrocarbon
systems, only 9 of the 122 points are outside the (25%
range. In general, pressure drop is underpredicted in
the near-flooding zone. A representative performance
plot showing the pressure drop fit is given in Figure 10.

Figure 10. Example comparison of predicted and experimental


pressure drops for dual-flow trays. Cyclohexane/n-heptane, total
reflux, 1.63 atm. Column diameter ) 1.2 m, tray spacing ) 0.91
m, open area ) 19%, hole diameter ) 12.7 mm. FRI data.

The efficiency curve has been included to show the


efficiency peak at the pressure drop curve break.
Froth Height. A two-phase porosity relationship for
dual-flow trays was developed by Mahendru and Hackl15
and used by Xu et al.8 to predict liquid holdup for a 300mm (11.8-in.) column operated with the methanol/water

Ind. Eng. Chem. Res., Vol. 41, No. 6, 2002 1637

and methanol/2-propanol systems. The porosity  (volume fraction of vapor in the froth) determined by Xu et
al. is

 ) 1.0 - 0.0946

[ ]
Us2FG
ghLFL

-0.2

refa

(13)

Accordingly, the froth height for dual-flow trays, hf, can


be calculated from the porosity and equivalent clear
liquid height

hf )

hL
(1 - )

Table 2. Comparison of Calculated and Measured


Efficiency for Dual-Flow Trays

(14)

As a note of caution, eqs 10 and 11 should be used only


in the vicinity of the peak efficiency, i.e., where the froth
is well-developed and stable. At low loadings, the liquid
on the tray is not well-aerated.
Mass Transfer Efficiency
Spray Zone. Previous models (e.g., Xu et al.8) follow
earlier work with cross-flow trays and are based on the
assumption that essentially all of the interphase transfer occurs within the liquid-continuous froth immediately above the tray floor. However, as pointed out
earlier, there can be significant mass transfer in the
spray zone above the froth, where all of the downflow
liquid comes into contact with the rising vapor (thereby
differing from conventional cross-flow trays). Because
it is likely that complete horizontal mixing takes place
in the froth (in the region of peak efficiency), the
additional mass transfer should be taken into account.
Because the froth is well-mixed, efficiency in the froth
cannot exceed 100% and because FRI obtained efficiencies as high as 150% (Table 1), it is clear that, for some
systems, mass transfer in the spray zone can be
significant. The liquid flowing through this zone has
been observed to flow primarily as streams, with
breakup into droplets for low-surface-tension systems
such as i-butane/n-butane. The resulting additional
interfacial area is available for mass transfer.
There appear to be no published data for spray zone
mass transfer. Studies by the Separations Research
Program have included empty-column contacting of
rising vapor with liquid fed through a conventional
distributor (430 streams/m2) normally used for packed
column work. A maximum of one theoretical stage was
obtained at contacting heights in the range of 3.5 m.
Clearly, for representative heights of the spray zone in
dual-flow trays (less than a tray spacing), one might
expect no more than 0.1-0.2 theoretical stages, depending on conditions. At present, any correction with a mass
transfer model to account for spray zone transfer must
be regarded as empirical. Analysis of the FRI data
indicates that up to 0.2 theoretical stages can be
achieved in the spray zone, depending on the system
and flow conditions. It is prudent, considering the
present level of knowledge, to neglect the mass transfer
contribution of the spray zone.
Froth Zone. Sieve tray mass transfer in the froth
zone has been modeled successfully for cross-flow sieve
trays. If we assume that mass transfer behavior in a
dual-flow froth can be treated the same as in a crossflow froth, as mentioned earlier, then we can estimate
the value of the dual-flow efficiency using models
developed for cross-flow sieve trays. The point efficiency

FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI
FRI

system

C6/C7
C6/C7
C6/C7
C6/C7
C6/C7
C6/C7
C6/C7
C6/C7
C6/C7
C6/C7
C4
C4
C4
C4
C4
IPA/H2O
xylenes
xylenes
C8/C10
alcohols
FRI C8/C10
alcohols
5
MeOH/H2O
FRI MeOH/H2O
2
Bz/Tol

hole
efficiency,
pressure TS
% diam peak, calculated
(kPa)
(m) open (mm) meas
(%)b
28
28
28
166
166
166
166
166
345
345
1138
1138
1138
1138
1138
101
16 mm
16 mm
10 mm

0.61
0.61
0.61
0.61
0.61
0.91
0.61
0.61
0.61
0.91
0.61
0.30
0.61
0.61
0.91
0.61
0.61
0.61
0.61

13.5
18.9
25.6
13.5
19.1
19.1
18.9
25.6
18.9
29.3
13.5
19.1
18.9
25.6
29.3
19.1
12.9
17.8
12.9

12.7
25.4
25.4
12.7
12.7
12.7
25.4
25.4
25.4
11.9
12.7
25.4
25.4
25.4
11.0
12.7
12.7
12.7
12.7

79
75
63
83
81
57
81
55
98
67
121
90
115
92
96
75
81
63
72

43
53
54
67
69
71
74
70
75
71
77
71
77
75
74
69
60
62
72

10 mm

0.61 17.8

12.7

42

86

101
101
101

0.41 14.2
0.41 18.2
0.51 14

c
c
12.7

82
82
87

80
80
60

a Column diameters: FRI, 1.2 m; ref 5, 1.0 m; ref 2, 0.45 m.


Contribution of spray zone not included. c Slots, 4 mm 150 mm.

model of Garcia and Fair16 was selected as the most


recent and extensively tested sieve tray model available.
To employ the model, one must utilize a column
diameter such that the active (bubbling) area is equivalent (after considering downcomer area) to the total
cross-sectional area of the dual-flow tray.
The correspondence between modeled and measured
peak efficiencies for each system is shown in Table 2.
In almost all cases, the predicted value is lower than
the observed value, lending some credence to the
importance of mass transfer in the spray zone. If a
multiplier of 1.2 is used to account for mass transfer in
the spray zone, better agreement is obtained, as shown
in Figure 11. This multiplier is empirical, based on the
earlier discussion, and the modeled efficiency for the
froth zone is more fundamental, although more conservative. Clearly, the model prediction must be discounted
at lower loadings. The low-load correction is empirical,
based on experimental observations. The ratio of corrected efficiency to peak efficiency can be obtained
through a rearrangement of eq 3

E
)
Ep

1 + Ep
1 -

(15)

Values of calculated from experimental data are


shown in Figure 12. The scatter indicates that this
simplistic approach might neglect some geometric and
mechanistic effects. However, prudent tray specifications call for hole sizes in the 12.7-25.4-mm range and
open areas in the 15-20% range. The lines in the
summary plot (Figure 12d) can be adapted to most
practical situations.
Tray Geometry Variables
In the present study, emphasis has been placed on
dual-flow trays with 15-20% open area, 12.7-mm holes,

1638

Ind. Eng. Chem. Res., Vol. 41, No. 6, 2002

Figure 11. Parity plot, predicted vs observed peak efficiency for


dual-flow trays.

and tray spacings of 0.5-0.7 m. The reason is that


design studies have shown that, for most applications,
this is an economical combination of tray geometric
variables. For example, Figure 5 shows the capacity
advantage of large open area and high tray spacing.
However, when mass transfer efficiency and pressure
drop are added to the comparison, the extreme cases of
geometry might not be optimum. Still, this is a matter
for detailed analysis of individual cases.
Figure 7 shows the effect of hole diameter on efficiency and pressure drop for the cyclohexane/n-heptane system at 1.63 atm and 19.1% open area. The
larger holes have a higher efficiency but also a higher
pressure drop. The effect of tray spacing on pressure
drop is minor if operation is well below the flood point.
Tray spacing can affect the efficiency in that it can
provide extra volume for mass transfer in the spray
zone.
Finally, mention should be made of tray diameter
effects. Figure 3 shows a difference between 0.45- and
2.50-m-diameter trays, more in capacity than in efficiency. The importance of maintaining a good liquid

Figure 12. Low-loading discount factor for representative experimental data: (a) 12.7-mm holes, 18-20% open area; (b) 25.4-mm
holes, 18-20% open area; (c) 12.7-mm holes or smaller, 13-15% open area; (d) summary plot.

Ind. Eng. Chem. Res., Vol. 41, No. 6, 2002 1639

distribution has been mentioned (spray nozzles are often


used to add reflux to the top tray). Zuiderweg et al.2
observed generally higher capacities and lower pressure
drops for larger tray diameters, suggesting that the
methods given here, largely based on a column diameter
of 1.2 m, are somewhat conservative. However, reports
of poor dual-flow tray performance for larger diameters
and very low flow parameters (as in high-vacuum
columns) must be recognized. It is possible that very
high volumetric ratios of vapor to liquid create a
tendency for the vapor flow to tend toward the center
portion of the tray, leading to reduced tray efficiency.
Thus, the findings described in this paper should be
used with great caution for the low values of the flow
parameter LML/VMVxFG/FL.
Summary and Conclusions
We have proposed a rational method for the analysis
and design of dual-flow tray distillation columns. The
approach is based on a large number of performance
tests under distillation conditions in a large-diameter
column. No rational model useful for commercial-scale
dual-flow tray column design or analysis has been
published previously. We have made suggestions for
possible mechanisms of phase contacting in the liquidcontinuous froth as well as in the vapor-continuous
spray zone. There is a clear need for additional studies,
both experimental and theoretical.
The following sequence of steps should be followed in
designing or rating dual-flow columns: (1) The column
diameter should be determined on the basis of flooding
limits and a prudent approach to flooding. (This is the
same procedure used for other contacting devices such
as cross-flow trays and packings.) An approach to
flooding of about 75-80% is in the range of peak
efficiency (Table 1). The effect of entrained liquid in the
vapor can be used to pinpoint the flooding approach as
well as the peak efficiency. (2) The peak efficiency
should be modeled by an adaptation of a froth-contacting
model for cross-flow sieve trays. To use the model, one
needs to know the fraction of holes that are active.
Because the model provides only the efficiency in the
froth, enhancement of this efficiency can occur through
added mass transfer above the froth (spray zone). On
the basis of the FRI measurements, the enhanced
efficiency can be as much as 1.2 times the modeled
efficiency, but until more definitive studies are made,
a factor of 1.0 is recommended. (3) The efficiency below
the peak point should be obtained by an empirical
correction to the peak efficiency. Tentative plots have
been provided to enable an estimation of the effect of
loading on efficiency. (4) The pressure drop should be
calculated on the basis of conventional equations, taking
into account the fraction of the total holes passing vapor
and the porosity of the froth.
This is a pioneering attempt at the modeling of dualflow tray performance. Future studies under distillation
conditions might require experimental techniques not
yet fully developed, e.g., X-ray tomography to show froth
quality changes and the dynamic activity at the holes.
Because such work has not yet been effective for the
more conventional devices, one might not expect dualflow technology advances soon. Perhaps the work reported here will help reactivate interest in a device that
is efficient and inexpensivesand that can operate well
under fouling situations.

Acknowledgment
This paper would not have been possible without the
generous provision of large-scale performance data by
Fractionation Research, Inc. (FRI). The authors extend
thanks to FRI, as well as to the Separations Research
Program at the University of Texas and to Koch-Glitsch,
Inc., for providing support for the modeling and analysis
effort.
Nomenclature
A, B, C ) constants in eq 12
Aa ) tray active area, m2
Ah ) tray hole area, m2
At ) total column cross-sectional (superficial) area, m2
Ax ) area of holes open to vapor at any instant ) Ah, m2
b1, b2, b3 ) constants in eq 8
Csb ) Souders-Brown capacity parameter, m/s
Cv ) orifice coefficient, eq 6
dh ) hole diameter, mm or m
Ep ) peak efficiency, fractional
Ew ) efficiency corrected for liquid-in-vapor entrainment,
fractional
Fs ) vapor F factor based on superficial area, UsFG0.5, m/s
(kg/m3)0.5
g ) gravitational constant, m/s2
h ) pressure loss, m of tray liquid
hdG ) drop for vapor flow through orifices
hL ) residual drop through two-phase mixture
hLD ) drop for liquid flow through orifices
hL ) head required to force liquid through the orifices
ht ) total pressure drop for a dual-flow tray
ht,meas ) measured total pressure drop for a dual-flow tray
hf ) height of froth on the tray, m
L ) liquid flow, kg mol/s
m ) slope of the equilibrium curve
n ) exponent in eq 8
ML, MG ) molecular weights of liquid and vapor, respectively
Q ) volumetric flow rate of vapor, m3/s
tt ) tray metal thickness, mm
TS ) tray spacing, m
U ) vapor velocity, m/s
Us ) superficial velocity
Ux ) velocity through x fraction of total number of holes
V ) vapor flow, kg mol/s
) fraction of total holes passing vapor at any instant
Greek Letters
 ) void fraction in the froth
) ratio of slopes, equilibrium curve to operating line )
mV/L
FL ) liquid density, kg/m3
FG ) vapor density, kg/m3
) ratio of wet efficiency (with entrainment) to dry (peak)
efficiency
) ratio of lower loading efficiency to peak efficiency
) efficiency discount factor for entrainment in vapor
) efficiency discount factor for lower loadings

Literature Cited
(1) Zuiderweg, F. J.; Verburg, H.; Gilissen, F. A. H. Comparison
of fractionating devices. Proc. Int. Symp. Distill. 1960, 201.
(2) Zuiderweg, F. J.; de Groot, J. H.; Meeboer, B.; van der Meer,
D. Scaling up distillation plates. Proc. Int. Symp. Distill. 1969, 5,
78.
(3) Huml, M.; Standart, G. The hydraulics of large turbogrid
trays. Brit. Chem. Eng. 1966, 11 (11), 1370.

1640

Ind. Eng. Chem. Res., Vol. 41, No. 6, 2002

(4) Kastanek, F.; Huml, M.; Braun, V. Measuring the efficiency


of a column of one metre diameter. Proc. Int. Symp. Distill. 1969,
5, 100.
(5) Kastanek, F.; Rylek, M. Turbogrid tray efficiencies. Collect.
Czech. Chem. Commun. 1970, 35, 3367.
(6) Kastanek, F.; Standart, G. Efficiency of selected types of
large distillation trays at total reflux. Sep. Sci. 1967, 2, 439.
(7) Fractionation Research, Inc., Stillwater, Oklahoma. Research Progress Reports, 1955-1958. Available from Oklahoma
State University.
(8) Xu, Z. P.; Afacan, A.; Chuang, K. T. Efficiency of dualflow
trays in distillation. Can. J. Chem. Eng. 1994, 72, 607.
(9) Baird, J. L. Letter to the editor. Chem. Eng. Prog. 1999, 95
(5), 7.
(10) Fair, J. R. How to predict sieve tray entrainment and
flooding. Petro/Chem Eng. 1961, 33 (10), 45. Also, recent editions
of Perrys Chemical Engineers Handbook; McGraw-Hill: New
York.
(11) Colburn, A. P. Effect of entrainment on plate efficiency in
distillation. Ind. Eng. Chem. 1936, 28, 536.

(12) Fair, J. R. Entrainment-efficiency effects on distillation


sieve trays. Presented at the AIChE Annual Meeting, Chicago,
IL, Nov 15, 1996.
(13) Leibson, I.; Kelley, R. E.; Bullington, L. A. How to design
perforated trays. Pet. Ref. 1957, 36 (2), 127.
(14) Kotschering, N. A.; Oleski, W. M.; Dilman, W. W. Investigation of dualflow tray behavior under distillation conditions.
Khim. Prom. 1960, 7, 591.
(15) Mahendru, H. L.; Hackl, A. Contribution to the design of
sieve trays without downcomers. Inst. Chem. Eng. Symp. Ser. 56
1979, 3.2/35-47.
(16) Garcia, J. A.; Fair, J. R. A fundamental model for the
prediction of distillation sieve tray efficiency. Ind. Eng. Chem. Res.
2000, 39, 1809 (part 1), 1818 (part 2).

Received for review April 12, 2001


Revised manuscript received December 13, 2001
Accepted December 17, 2001
IE010326W

You might also like