DISTILLATION TRAYS THAT OPERATE

BEYOND THE LIMITS OF GRAVITY BY

USING CENTRIFUGAL SEPARATION

P. Wilkinson1,
, E. Vos1, G. Konijn1, H. Kooijman1, G. Mosca2 and

L. Tonon2

1
Shell Global Solutions, Amsterdam, The Netherlands.
2
Sulzer Chemtech, Winterthur, Switzerland.

Abstract: The interest of industry in super-high capacity fractionation trays has significantly
increased in the last few years (Bravo and Kusters, 2000). This paper focuses on a comparison
of these technologies, using available data from open literature. In addition new research data for
the ConSep tray will be presented using state of the art gamma-scanning capabilities. Four tray
technologies fall in this category—The shell swirltube tray, Jaeger COFLO tray, Koch-Glitsch
ULTRA-FRAC tray and the shell ConSep tray. Publications on these trays have focused on a
broad spectrum of applications—low liquid load operations such as glycol contractors (Weiland
and Griesel, 2004) and low-pressure applications such as a hydrocracker main fractionator
(Groenendaal et al., 2001) to high-pressure systems such as refinery superfractionators
(De Villiers et al., 2005; Mosca et al., 2004a, b), debutanizers (De Villiers et al., 2004, 2005)
and depropanizers (Wilkinson et al., 2006).
Keywords: trays; high-performance; capacity; system limit.

The interest of industry in super-high capacity occur. As the pressure increases the density
fractionation trays has significantly increased difference between the phases decreases
in the last few years (Bravo and Kusters, and the velocities required to reach the so-
2000). This paper focuses on a comparison of called ‘system limit’ get lower and lower. In
these technologies, using available data from this paper we have used the correlations
open literature. In addition new research data given by Stupin and Kister (2003) to calculate


Correspondence to: for the ConSep tray will be presented using the system limit conditions. The system limit
Dr. P. Wilkinson, Shell Global
Solutions, Badhuisweg 3,
state of the art gamma-scanning capabilities. calculated in this way will be used in this
Amsterdam 1031, CM, Four tray technologies fall in this category– paper to assess to what extent novel tray
The Netherlands. The Shell Swirltube tray, Jaeger COFLO tray, technologies can exceed the system limit.
E-mail: peter.wilkinson@ Koch-Glitsch ULTRA-FRAC tray and the Shell The technologies assessed in this paper are:
shell.com
ConSep tray. Publications on these trays Jaeger COFLO tray (Fair et al., 1999), Koch-
DOI: 10.1205/cherd06103 have focused on a broad spectrum of appli- Glitsch ULTRA-FRAC tray (Weiland and
cations—low liquid load operation such as Griesel, 2004; Weiland and Resetarits, 2002;
0263–8762/07/ glycol contactors (Weiland and Griesel, 2004) Bulletin KGUF-13M09/01B, 2001) and Shell
$30.00 þ 0.00 and low-pressure applications such as a hydro- ConSep tray (Groenendaal et al., 2001;
Chemical Engineering
cracker main fractionator (Groenendaal et al., Mosca et al., 2004a). These trays have all
Research and Design 2001) to high-pressure systems such as refin- used some kind of separator device to prevent
ery Superfractionators (De Villiers et al., 2005; excessive entrainment as shown below in
Trans IChemE, Mosca et al., 2004a, b) debutanizers (De Table 1.
Part A, January 2007
Villiers et al., 2004, 2005) and depropanizers Clearly these trays differ in the type of contact
# 2007 Institution (Wilkinson et al., 2006). and separation devices used. Therefore com-
of Chemical Engineers paring and evaluating these different internals
LIMITATIONS OF ‘CONVENTIONAL· requires a closer look at the performance of
HIGH CAPACITY TRAYS the de-entrainment devices used in these trays.

The term ‘system limit’ is often used as the

GAS -- LIQUID SEPARATORS
point above which vapour and liquid velocities
(depending on physical properties) become so For use in distillation trays the separator
high that excessive entrainment will always should have the following characteristics:

130 Vol 85 (A1) 130–135

 DISTILLATION TRAYS THAT OPERATE BEYOND LIMITS OF GRAVITY 131

Table 1. Various types of super-high capacity trays and their differences.

Tray type Contact and separation zones Contact type De-entrainment internal

Jaeger COFLO tray Separate Normal distillation tray Baffle-type entrainment collector
Shell Swirltube tray Combined Centrifugal contact Swirl tube separator
Koch-Glitsch ULTRA-FRAC tray Combined Centrifugal contact Centrifugal separator
Shell ConSep tray Separate Normal distillation tray Swirl tube separator

. High vapour/liquid separation efficiency (.95% to prevent correlation for the system limit is used which has been
substantial tray efficiency loss). published by Stupin and Kister (2003):
. Good turn-down performance.
. Low cost. C1 ¼ 0:445
(1  F)(s=Dr)0:25  1:4Ls
. Low pressure drop.
. A good ‘mass-transfer efficiency’ will also be essential for which can be rewritten to:
those devices where contacting and separation are com-
bined in a single element. 0:445
(1  F)(s=Dr)0:25
C1 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi
1 þ 1:4
f
Dr=rL
The baffle-type entrainment collector of the COFLO tray and
the swirl tube separator in the Shell ConSep tray can be pffiffiffiffiffiffiffiffiffiffiffi
using the flow-parameter ¼ w ¼ Ml =Mg rg =rl
traced back to typical gas–liquid separators commonly used
in industry i.e., the vane-pack mist-eliminators and the cyclonic C2 ¼ 0:356
(1  F)(s=Dr)0:25
separators. Although both of these devices have high de-
entrainment efficiency, they differ in performance in terms of and
the maximum vapour and liquid loads that they can handle.
While the vane-pack and cyclonic separators can handle 1
F¼
high vapour loads at low flow parameters the performance 1 þ 1:4(Dr=rg )0:5
changes markedly at high pressures and/or increased liquid
loads (Robinson, 2001, 2003; Perry, 1984). There are two where CS,ult ¼ the smallest between C1 and C2.
reasons for this. Firstly, vane packs generate the centrifugal Results for the C6/C7 system at different pressures are
forces (‘g’ forces) through the oscillatory gas flow path shown in Figure 1 below. This figure confirms that the
between the plates or baffles whereas the cyclonic separators COFLO tray can operate above the system limit provided
use swirlers to impart strong axial and radial flows to the two- the liquid loads are relatively low (in this case below or
phase flow. Because of the use of swirlers, cyclonic separators close to a liquid load of 36 m3 m2 h21). The rapid decline in
generate higher ‘g’ forces as compared to the vane packs and capacity as the pressure is raised is most likely due to limit-
hence have higher capacity. Secondly, at higher liquid loads ations in the liquid handling capacity of the applied separator
re-entrainment of liquid from the vane surfaces becomes system.
predominant and hence their vapour handling capacity For the ConSep tray the swirlers have a typical maximum
decreases dramatically. Cyclonic separators on the other liquid handling capacity in the order of:
hand can handle higher liquid loads quite well. Based on
these considerations it is expected that the COFLO tray will LmaxSwirler  1100 m3 m2 h1
be capable of operating beyond the system limit provided the
liquid loads are not too high. In order to assess this point a (based on swirler cross sectional area).

Figure 1. COFLO (Fair et al., 1999) flood points for C6/C7 at three different pressures compared with the system limit correlation.

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A1): 130–135
132 WILKINSON et al.

Figure 2. ConSep air– water data (measured in a 45 cm ID column) for a layout with 18% swirltube area.

While the maximum vapour capacity of the swirlers is (Weiland and Griesel, 2004). Obviously this is an additional
advantage for these kind of trays since the liquid handling
Cmaxswirler  1 m s1 capacity of the downcomer is improved when clear liquid is
entering the downcomer instead of an ‘aerated’ liquid.
(C-factor based on swirler cross sectional area). Figure 4 shows hydrocarbon data measured with ConSep
Due to geometrical constraints (depending on area in one of our research test facilities in Amsterdam (together
required for downcomers and so on) only about 20– 30% of with an FRI data-point which is in good agreement with our
the column cross sectional area will be occupied by swirlers own results for iC4/nC4 which shows that the effect of scale
and therefore the liquid handling capacity of the ConSep 1.2 m ID for FRI versus 0.45 m ID measured at Shell is
separators will be about Lmaxcolumn ¼ Lmax swirler * %Swirler negligible). The swirl-tube area for the C4 data was 18%.
cross-sectional area/100%  200– 320 m3 m22 h21 based Important conclusions that can be drawn from Figure 4 are:
on total column cross sectional area. Judging from Figure 1
. Maximum operating points are well above the system limit
this is substantially higher than for the COFLO tray.
for high as well as low flow parameters.
Experimental results for the air –water system at ambient
. The maximum operating points remain below the maxi-
conditions with a ConSep tray having 18% swirltube area mum (separator determined) values shown in Figure 2.
are shown in Figure 2.
This is due to the fact that the capacity for these layouts
The distance between the sieve deck and the separator
are constrained by downcomer back-up.
deck was varied between 200 and 600 mm. The results in
Figure 2 are for a (vertical) sieve deck to separator deck dis-
tance of 400 mm (¼height of froth contacting zone on sieve
deck). A special feature of this ConSep test facility was that
downcomer back-up limitations were prevented by using a
very long downcomer. Under these conditions the maximum
capacities achieved for the ConSep tray are clearly well
described by the separator performance equations. Clearly
both vapour and liquid loads are very high and illustrate the
fact that the separators applied in ConSep will normally not
pose any constraint. In fact the normal upper limit for a
ConSep tray will be determined by downcomer back-up.
Most of the liquid entering the downcomer has been
‘degassed’ in the separators and therefore a high froth den-
sity is typically achieved in the downcomer. This has been
confirmed by visual observations for the air/water measure-
ments. In addition detailed gamma-scans (Figure 3) show
that the density at the bottom of the downcomers under
hydrocarbon conditions also show a very high liquid content. Figure 3. Gamma-scans for iC4/nC4 mixture at 11 bar, tray
spacing ¼ 450 mm under total reflux conditions. Points in picture
The fact that the liquid (entering the downcomer) is very on the left show ‘scan points’. The vapour load shown is for
much degassed by the separator has also been reported Cs ¼ 0.12 m s21 near flooding conditions (Cs ¼ capacity factor
for the COFLO (Fair et al., 1999) and ULTRA-FRAC tray based on total column cross-sectional area).

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A1): 130– 135
 DISTILLATION TRAYS THAT OPERATE BEYOND LIMITS OF GRAVITY 133

Figure 4. ConSep hydrocarbon data measured in a 45 cm ID column (iC4/nC4 11 bar circulation and total reflux runs þC6/C7 140 kPa total
reflux runs). The FRI data point was measured in a 1.2 m ID column.

Since the ULTRA-FRAC tray also contains an internal Furthermore, the data show that liquid handling capacity of
using centrifugal force for the vapour/liquid separation it is these trays is indeed higher than for the COFLO tray as
expected that this internal should also be able to operate expected for a device where the separator is based on cen-
beyond the system limit. Some air–water data for ULTRA- trifugal forces. However, for the higher flow parameters the
FRAC has been published as shown in Figure 5 confirming air –water data remain below the system limit. In addition
that the maximum operating conditions at low liquid flow it was reported in 2002 by Koch-Glitsch (Weiland and
rates indeed operate above the system limit for air –water. Resetarits, 2002) that the ULTRA-FRAC tray does not

Figure 5. Maximum (air/water) capacity data for ULTRA-FRAC taken from Weiland and Griesel (2004) and Bulletin KGUF-1 3M09/01B (2001).

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A1): 130–135
134 WILKINSON et al.

Table 2. Prediction of tray efficiency with the O’Connell correlation for super-high capacity trays and comparison with observed efficiency.

Efficiency

Tray Application Actual (%) O’Connell predicted (%)

ConSep Tray (De Villiers et al., 2004) IC4/nC4 at 11 bar 89 82

COFLO tray (Fair et al., 1999) C6/C7 at 0.33 bar 60 55
C6/C7 at 1.06 bar 70 58
C6/C7 at 1.66 bar 75 63
ULTRA-FRAC tray (Bulletin KGUF-1 3M09/01B, 2001) Deethanizer 85 82
Depropanizer 78– 82 79
Debutanizer 75– 85 69

exceed the system limit flood under high pressure conditions. In order to compare the data we have taken the same
The authors say, ‘These trays might be thought to be capable approach by comparing the O’Connell correlation predicted
of exceeding system limit flood. However, this proves not to efficiencies with the measured efficiencies for the different
be the case.’ So what makes efficient gas–liquid separation super-high capacity trays. The data (Table 2) confirm that
at high liquid loads and high pressures so difficult? The the O’Connel correlation provides a reasonably (conserva-
answer probably lies in the ‘g-forces’ one is able to impart tive) estimate for the efficiencies.
to the gas–liquid mixture once it leaves the tray. This trans-
lates into centrifugal acceleration, C.
CONCLUSIONS
V 2r Several super-high capacity trays are now available on the
C¼ (m s2 )
r market. For low pressure/low liquid loaded systems all of
these products have shown evidence for achieving capacities
where Vr ¼ radial velocity (m s21) and r ¼ radius of travel (m). exceeding the system limit. This is clearly a breakthrough in
Centrifugal separators provide the radial velocity via swir- distillation technology. For higher liquid loaded (often high
lers present at the inlet side. Exact swirler dimension have pressure) systems it appears that the use of a centrifugal
not been published for ULTRA-FRAC or ConSep. However, type separator is required to achieve large capacity gains
based on pictures in the open literature (De Villiers et al., beyond the system limit. Of the different trays evaluated in
2004; Weiland et al., 2005) it is obvious that the diameter this paper the COFLO is the only one not using a centrifugal
of the ConSep swirlers are probably a factor 2–3 smaller type separator, and therefore it is to be expected that this tray
than used for ULTRA-FRAC which contributes to high will not exceed the system limit at higher pressures and/or
g-forces and improved separation of ConSep swirlers under higher liquid loads. The data reported for COFLO (Fair et al.,
high pressure conditions. 1999) support this assumption as operating beyond the
system limit has only been demonstrated under low pressure
TRAY EFFICIENCY conditions for this device. When comparing ConSep to
ULTRA-FRAC it appears that both have a higher liquid
For ‘conventional’ trays with long flow-path lengths tray effi- handling capacity than COFLO and are also more suitable
ciencies well above a 100% are often reported. Values above for operating at high pressures. The use of smaller swirl
100% are mostly the result of some degree of staging. For the elements in ConSep should contribute to higher g-forces as
super-high capacity trays discussed in this paper staging compared to ULTRA-FRAC. Possibly this makes ConSep
effects are likely to be more or less absent. Consequently it trays more suitable for operating under very high pressure
is expected that the maximum attainable efficiency for conditions when gas-liquid separation becomes more difficult.
these devices will always be below 100%.
For the ConSep tray and the COFLO tray it is to be
expected that the overall tray efficiency is determined by NOMENCLATURE
two contacting steps: Csubscript vapour load factor (Sauders– Braun velocity), m s21
C centrifugal acceleration in centrifugal device, m s22
. Contacting on the tray deck (should be at least the point Cs,ult ultimate capacity based on system limit
efficiency). correlation, m s21
. Additional contacting in the separator section (e.g., high Eoc predicted tray efficiency from the O’Connell (1946)
g-forces in ConSep). correlation
F density factor in system limit correlation
For the ULTRA-FRAC tray and the Shell Swirl-tube tray Lsubscript liquid load, m3 m22 h21
contacting is primarily taking place in a single separator/ Ls liquid load in system limit correlation, m3 m22 s21
Msubscript massflow, kg s21
contacting device. r radius of travel in centrifugal device, m
It has been reported (Weiland and Resetarits, 2002) that Vr radial velocity in centrifugal device, m s21
the O’Connell (1946) correlation leads to reasonably good
predictions for the ULTRA-FRAC trays. This simple corre- Greek symbols
lation relates the tray efficiency to liquid viscosity and relative a relative volatility
volatility and is based on the test data from 31 plant columns. mL dynamic viscosity, centipoise
rsubscript density, kg m23
s surface tension
E OC ¼ 0:492(mL a)0:245 w flow parameter

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A1): 130– 135
 DISTILLATION TRAYS THAT OPERATE BEYOND LIMITS OF GRAVITY 135

Subscripts Mosca, G., Tonon, L., Wilkinson, P.M. and Reich-Rohrwig, P., 2004b,
g gas phase Improve propylene production with high performance trays, ERTC
l liquid phase Petrochemical Conference, Vienna, 11– 13 October.
max maximum O’Connell, H.E., 1946, Trans AIChE, 42: 741.
ult ultimate Perry, H., Green, D.W. and Maloney, J.O., 1984, Perry’s Chemical
Engineers’ Handbook, 6th edition, Figures 18–116 and 18– 124
(McGraw-Hill Book Co., Singapore).
Robinson, P., 2001, Mist elimination, Hydrocarbon Engineering,
March: 76.
REFERENCES Robinson, P., 2003, Mist separation, Hydrocarbon Engineering, June:
80.
Bravo, J.L. and Kusters, K.A., 2000, Tray technology for the new Stupin, W.J. and Kister, H.Z., 2003, System limit: The
millennium, Chem Eng Prog, December, 33– 37. ultimate capacity of fractionators, Trans IChemE, 81(Part A):
Bulletin KGUF-1 3M09/01B, Printed in U.S.A. #Copyright 2001 136– 146.
Koch-Glitsch, LP. Weiland, R.H. and Resetarits, M.R., 2002, Finding new uses for old
De Villiers, W.E., Bravo, J.L., Wilkinson, P.M. and Summers, D.R., distillation equations, Petroleum Technology Quarterly (PTQ),
2004, Further advances in light hydrocarbon fractionation, PTQ 115–121.
Summer. Weiland, R.H. and Griesel, C.A., 2004, Increasing the capacity
De Villiers, W.E., Wilkinson, P.M. and Summers, D.R., 2005, Appli- of glycol contactors, AIChE Spring Meeting, New Orleans, LA,
cation of ultra high capacity ConSep trays for debottlenecking of April.
gas and ethylene plant fractionation trains, New Orleans AIChE Weiland, R., DeGarmo, J. and Nieuwoudt, I., 2005, Converting a
Spring Meeting. commercial column into a research tower, CEP.
Fair, J.R., Trutna, W.R. and Seibert, A.F., 1999, A new, ultracapacity Wilkinson, P.M., DeVilliers, W., Mosca, G. and Tonon, L., 2006,
trays for distillation columns, Trans IChemE, 77(Part A): 619–626. Achieve challenging targets in propylene yield using ultra
Groenendaal, C., Trautrims, B., Kusters, K.A. and Bravo, J.L., 2001, system limit fractionation trays, ERTC Petrochemical Conference,
The Shell ConSepTM tray technology provides unparallel distillation Dusseldorf, 9 –11 October.
capacity, Presented at the EFChE Conference, Bamberg,
Germany, April.
Mosca, G., Tonon, L., Efremov, D. and Wilkinson, P.M., 2004a,
Expand plant’s capacity with high performance trays, RPTC Petro- The manuscript was received 14 July 2006 and accepted for
chemical Technology Conference, Moscow, 10–14 September. publication after revision 7 November 2006.

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A1): 130–135