Professional Documents
Culture Documents
Effects of Design on
Tray Efficiency in
Commercial Towers
Henry Z. Kister
Fluor Corp. When designing a new distillation column
or revamping an existing one,
it is important to understand
the factors that affect overall tray efficiency.
D
espite decades of work on tray efficiency, it is still good characterization of tray efficiencies in commercial
unclear which design variables affect tray efficien- towers. This concept is, therefore, preferred by most indus-
cies of commercial columns and to what extent. trial practitioners. Alternative definitions of tray efficiency
Many of the efficiency studies have been contradictory, are more often employed to relate tray efficiency to mass-
and others relied on data derived from laboratory-scale transfer fundamentals, as discussed by Lockett (1).
towers that do not extend to commercial towers. Based on Overall column efficiency is normally obtained from
much experience gathered in the industry over the years, test data by matching a stage-to-stage calculation, usually
and a large amount of commercial-scale data recently by means of a commercial tower simulation, to test data.
released by Fractionation Research Inc. (FRI), it is now Based on the measured mass and component balances,
possible to get a clearer picture of the factors that affect boilup, reflux, temperatures, pressures and compositions,
overall tray efficiency. the number of stages in the calculation is varied until the
Previous work showed that errors in vapor-liquid equi- closest match to data is obtained. Good practices for
librium (VLE) data and reflux ratio measurements can have obtaining reliable test data and for correctly deriving
a major effect on tray efficiency calculations. Past work also overall column efficiencies from them are discussed in
showed that overall tray efficiency increases with lower vis- Refs. 2 and 3.
cosities and relative volatilities. This article presents an Reliability of field and test data. Deriving reliable data
analysis of FRI data and quantitatively defines the effects of from industrial operating towers is difficult (24); flowme-
tower geometry flow path length, fractional hole area, ter and analyzer errors present some of the biggest chal-
hole diameter, and weir height on overall tray efficiency lenges (4). Tabulations of measured efficiency data obtained
in commercial-scale fractionators. from operating columns are available (e.g., Ref. 5), but their
limitations and reliability are not always well-defined.
The pitfalls of obtaining An excellent source of commercial-scale test data is
tray efficiency from test data measurements by FRI. For many years, only limited FRI
Overall column efficiency. The overall column (or data have been published in the open literature, but recently
overall tray) efficiency, EOC, is the ratio of the number of much more data have been released (68). The FRI tests
theoretical stages required for the separation (not counting were conducted in a research facility that permitted close
the reboiler(s) and condenser(s)) to the number of trays in monitoring, using standard test systems whose physical
the tower. Since tray efficiencies may vary from one tower properties are well-known. This arrangement overcomes
section to another, the overall column efficiency concept is many of the limitations encountered in the testing of operat-
often applied separately to the stripping section and the rec- ing plant towers. Most of the data discussed here was
tifying section. It is easy to apply, and generally provides a obtained by FRI.
true
% Error in 50
Efficiency 48
= 1.1
50 apparent < true
Figure 1. Direct effects of errors in relative volatility on tray been reported to have a small effect on tray efficiency (12).
efficiency are often not fully appreciated. Reprinted with permis- For measurements at finite reflux ratios, the direct effect
sion from Ref. 9, Copyright 1992, McGraw-Hill Companies. of errors in relative volatility () on tray efficiency (depicted
in Figure 1) are compounded by an interaction with the
Accuracy of VLE correlations. Errors in the VLE rela- reflux ratio (R). Consider the case where apparent < true and
tionships are the result of inaccuracies in VLE test data, and test data at a finite reflux are analyzed to calculate tray effi-
are known only within limits based on the source of the ciency. Due to the volatility difference, Rmin,apparent > Rmin,true.
VLE information. This gives rise to uncertainty in both the Since the test was conducted at a fixed reflux flowrate,
VLE data and the resulting stage-to-stage calculations. (R/Rmin)apparent < (R/Rmin)true. A calculation based on the
Errors in relative volatility are the most under-appreciat- apparent R/Rmin will yield more theoretical stages than a cal-
ed factor affecting tray efficiency. Figure 1 shows the direct culation based on the true R/Rmin, and therefore, a higher
effect of such errors (911). At very low relative volatilities apparent efficiency than the true value. In the case illustrated
( < 1.2), small errors in VLE data have a huge impact on in Figure 2, the calculated efficiency was double the true
tray efficiency. For instance, at = 1.1, a 3% error gives a value due to a relatively small error in VLE.
tray efficiency 4050% higher than its true value. Since the The indirect effects added to the direct effects of Figure
accuracy of VLE data is seldom better than 23%, in low- 1 widen the gap between the true efficiency and the appar-
volatility systems tray efficiencies become meaningless ent efficiency. The indirect effects escalate exponentially as
unless accompanied by VLE data. Likewise, comparing minimum reflux is approached. Near minimum reflux, even
efficiencies for a low-volatility system derived by different small errors in R/Rmin induce huge errors in the number of
sources is misleading unless they use identical VLE data. stages and, therefore, in tray efficiency. Thus, efficiency
This applies to both trayed and packed towers. data obtained near minimum reflux are meaningless and
Figure 1 also reveals that errors in relative volatility are a potentially misleading.
problem only at low relative volatilities. For > 1.5 to 2.0, Accuracy of mass balances and reflux measurements.
VLE errors have small direct impact on tray efficiency. Errors in mass balance or reflux measurements affect the
Most efficiency test data reported in the literature were efficiencies derived from test data at finite reflux. This
obtained at total reflux. This is true of most of FRIs pub- includes errors in the physical properties (such as densities
lished data as well as most of the other data derived from and latent heats) that convert external tower mass-flow
test columns. The total-reflux test mode is often preferred, measurements into the internal molar flows used in the sim-
because measurements under total reflux do not suffer from ulation. When the measured apparent molar reflux ratio is
compounding of inaccuracies due to finite reflux ratios (the less than the true reflux ratio, a calculation based on the
indirect VLE effects, Figure 2), nor do they suffer from measured reflux ratio will give a higher apparent efficiency
inaccuracies in mass and energy balances. Reflux ratio has than the true efficiency. As with the indirect VLE effects, the
Figure 3. Uniform-efficiency plateaus demonstrate that vapor the tray geometry and test system, and (in the absence of
and liquid loads have little impact on efficiency in the normal hydraulic limits) is independent of vapor and liquid loads.
operating range (between excessive weeping and excessive
entrainment). Also shown is the small increase in efficiency with This uniform-efficiency plateau is a measure of the efficien-
pressure. FRI data, total reflux, DT = 1.2 m, S = 610 mm, cy of the system and tray geometry.
hw = 50.8 mm, dH = 12.7 mm. Reprinted with permission This uniform-efficiency plateau concept is applied here to
from Ref. 13, Copyright 1982, American Chemical Society. evaluate the effects of tray geometry on tray efficiency. Test
data reported by FRI and others are averaged into a single
effects of errors in mass balance or reflux measurements on smoothed value for each uniform-efficiency plateau. These
tray efficiency escalate exponentially as minimum reflux is smoothed tray efficiencies then provide comparative informa-
approached. In this regard, too, efficiency data obtained near tion on the effects of various geometrical variables. For
minimum reflux are meaningless and potentially misleading. instance, the plateaus in Figure 3 indicate efficiencies of 110%
for isobutanen-butane, 82% for cyclohexanen-heptane at
Uniform-efficiency plateaus 1.65 bar (24 psia), and 71% for cyclohexanen-heptane at
in tray efficiency vs. loads plots 0.3 bar (5 psia).
Figure 3 is a plot of overall tray efficiencies vs. percent of
flood for three test systems (13) at total reflux. The percent Effects of tray geometry on efficiency
of flood increases with higher vapor and liquid loads. Figure The literature contains numerous investigations and dis-
3 shows that vapor and liquid loads have a small effect on cussions on the effects of tray geometry on overall column
overall tray efficiency, as long as no capacity or hydraulic efficiency. References 1, 5 and 9 report the current state of
limits (flood, weep, channeling etc.) are encountered. the art. Despite the extensive past work, an understanding
The uniform-efficiency plateau on each curve of Figure of the effects of tray geometry on efficiency in commercial
3 indicates the range of operation where neither capacity columns remains elusive. Evaluating the recently released
nor hydraulic limits exist. The fall-off in efficiency at low FRI data together with the uniform-efficiency plateau con-
percent of flood is due to weeping, and that at high percent cept provides a clearer picture of the effects of tray geome-
of flood is due to entrainment as flooding is approached. try on the overall efficiency of commercial columns.
References 6 and 7 present a multitude of similar plots of Most of the FRI data used here were for sieve trays. The
overall tray efficiency vs. load, all of which exhibit efficien- other data, for conventional valve trays (both moving-valve
cy plateaus similar to those of Figure 3. Most of the efficien- and fixed-valve trays), appear to be comparable to the sieve
cy test data in Ref. 8 exhibit similar efficiency plateaus when tray data. As noted elsewhere (1, 5, 9), other than with
plotted against percent of flood at total reflux. regard to weeping behavior, sieve and valve trays are com-
Thus, for a given test system and tray geometry, a uni- parable, and a relationship applying to one type readily
form-efficiency plateau exists. This plateau is a function of extends to the other. Thus, the discussion presented here
(a) should apply to both as well. All FRI sieve-tray data dis-
140 cussed here were obtained for 1.7-mm-thick trays with the
120
rough edge of the holes facing the vapor flow.
Fractional hole area. FRIs tests in commercial-scale
100 towers (6, 13) show an increase in tray efficiency when
Tray Efficiency, %
80
When the fractional hole area increased from 5% to 8%,
60
the high-pressure butane system (Figure 4a) experienced a
small increase in efficiency, the atmospheric
40 5% cyclohexanen-heptane system (Figure 4b) experienced lit-
8%
tle change in efficiency, and the vacuum cyclohexanen-
14%
20 heptane system (Figure 4c) experienced a significant reduc-
20%
0
tion in efficiency. Going from a fractional hole area of 8%
0.000 0.020 0.040 0.060 0.080 0.100 0.120 to 14% resulted in tray efficiency reductions in all test sys-
Capacity Factor, Cs, m/s tems, ranging from 10% for isobutanen-butane to a much
larger 1520% for cyclohexanen-heptane. (These results
(c) for cyclohexanen-heptane appear significantly higher than
those reported earlier by FRI (13).) Note that the data in
100 Figure 4 for a 20% hole area are not directly comparable to
90
the others, as they were measured for 0-mm (i.e., no weir)
80 and 150-mm-tall outlet weirs, whereas the other data in
Tray Efficiency, %
Af = 13.7%
100
120 Af = 8.4%
EOC, Overall Column Efficiency, %
Af = 13.7%
100 80
Af = 8.4%
80
Af = 13.7%
60
60
Isobutanen-Butane, 11.4 bar, 8.4% Hole Area
Sieve 8.4% Hole Area 13.7% Hole Area
40 Isobutanen-Butane, 11.4 bar
40 Cyclohexanen-Heptane, 1.65 bar
Cyclohexanen-Heptane, 1.65 bar 8.4% Hole Area 13.7% Hole Area
20 Cyclohexanen-Heptane, 0.3 bar Cyclohexanen-Heptane, 0.3 bar
8.4% Hole Area 13.7% Hole Area
0
0 200 400 600 800 1,000 0 50 100
Tray Spacing, mm Weir Height, mm
Figure. 6. Uniform-efficiency plateaus show the effect of tray Figure 7. Uniform-efficiency plateaus show the effect
spacing on tray efficiency at total reflux, with DT = 1.2 m, of weir height on tray efficiency at total reflux, with
Af = 8.4%, hw = 50.8 mm, dH = 12.7 mm, and FPL = 760 mm. DT = 1.2 m, S = 610 mm, Af = 8.4% and 13.7%,
Source: FRI data. dH = 12.7 mm, and FPL = 760 mm. Source: FRI data.
100
EOC, Overall Column Efficiency, %
60 100
Isobutanen-butane
40
11.4 bar
Cyclohexanen-Heptane, 1.65 bar 80 20.7 bar
Cyclohexanen-Heptane, 0.3 bar 27.6 bar
20
IPA-Water, 1 bar
0
250 500 750 1,000 1,250 1,500 250 500 750 1,000 1,250 1,500
Flow Path Length, mm Flow Path Length, mm
Inlet Inlet
120
21
EOC = 0.492(L)0.245 (2) Figure 12. Uniform-efficiency plateaus show the effect
of pressure (at high presures) on tray efficiency for
isobutanen-butane; all operated at total reflux, with
Further theoretical validation of these parameters was provid- DT = 1.2 m, Af = 8.4%, dH = 12.7 mm, and hw = 50.8 mm.
ed by Chen and Chuang (30), who showed that the OConnell Red curve: FPL = 760 mm, 13% (of tower cross-sectional area)
correlation can be derived from theory if one assumes that dis- straight downcomer, S = 610 mm.
tillation mass transfer is controlled by the liquid phase. Green curve: FPL = 610 mm, sloped (30% top/7.5% bottom)
downcomer, S = 610 mm.
Pressure. Tray efficiency slightly increases with pressure Blue curve: FPL = 610 mm, sloped (30% top/7.5% bottom)
(Figure 3), reflecting the increase in efficiency associated downcomer, S = 914 mm. Source: FRI data.
with the reductions in liquid viscosity and in relative volatili-
ty that generally accompany a distillation pressure increase. comer geometries and different tray spacings. The efficiency
At pressures exceeding 10 bar, especially at high liquid decrease is of the same order for the different geometries.
flowrates, vapor entrainment into the downcomer liquid Surface tension. The literature contains extensive dis-
becomes important, and tray efficiency decreases with fur- cussion regarding the effect of surface tension on tray effi-
ther increases in pressure (31). This is well-illustrated in ciency, but no consensus has been reached. More-detailed
Figures 10 and 12 using FRI data for the isobutanen-butane discussions are available elsewhere (1, 9, 32). Often, several
system at pressures from 11.4 bar to 27.6 bar. The decrease physical properties vary from one test system to another, so
in efficiency over this pressure range is on the order of it is difficult to divorce the surface tension effects from
1520%. The curves in Figure 11 represent different down- those of other physical properties. CEP
HIGHLIGHTS
ast work showed that errors in the VLE correlations have a Flow path length is the primary geometric factor affecting
Pmajor effect on measured and calculated tray efficiency for low- tray efficiency. Increasing the number of passes from one to two
volatility systems and in those operated close to minimum reflux. or from two to four can reduce tray efficiency by as much as
Efficiencies for low-volatility systems are meaningless unless 1020% at lower pressures (0.3 to 1.7 bar) and by 515% at
accompanied by the VLE data. Efficiency data obtained near mini- higher pressures (1128 bar).
mum reflux are also meaningless and potentially misleading. Increasing the fractional hole area on trays from 5% to 8%
Past work also shows that lower liquid viscosities and lower of the active area has a minor effect on tray efficiency, but a fur-
relative volatilities improve overall tray efficiencies. As distillation ther increase to 14% reduces tray efficiency by 510%. At the
pressure increases, tray efficiency first rises due to the lower vis- larger hole areas, there may also be an increased sensitivity to
cosity and relative volatility. At pressures exceeding 10 bar, how- leakage at the support rings.
ever, efficiency declines with a rise in tower pressure, probably Changes in tray spacing have a minor effect on tray efficiency.
due to vapor recycle in the downcomers. As hole diameter increases from 3 mm to 13 mm, tray
Our data survey found that in the absence of hydraulic limita- efficiency increases by 510%. At diameters between 13 mm and
tions, such as weeping, entrainment or channeling, the percent of 25 mm, tray efficiency remains constant. Further increases in hole
flood has little effect on tray efficiency. On this basis, we define the diameter cause the tray efficiency to decline.
uniform-efficiency plateau, which is the flat portion of the overall At fractional hole areas of 8-9%, taller weirs have a minor
tray efficiency vs. percent of flood plot. This uniform-efficiency effect on tray efficiency. At hole areas around 14%, taller weirs
plateau is a measure of the efficiency of the system and tray may slightly enhance efficiency.
geometry in the absence of hydraulic issues, and was used to Experience and recent literature show that maldistribution
compare test data for various systems and tray geometries. in multi-pass trays may lead to further major reductions in tray
The current survey and data analysis identified the main geo- efficiency.
metric variables affecting tray efficiency:
HENRY Z. KISTER is a Fluor Corp. senior fellow and director of fractionation Acknowledgment
technology at Fluor Corp. (Phone: (949) 349-4679, E-mail:
The author is grateful to Fluors Walter Stupin and Matthew Olsson and
henry.kister@fluor.com). He has over 30 years of experience in design,
to consultant (and former FRI expert) Tak Yanagi for reviewing this work
troubleshooting, revamping, field consulting, control and startup of
and for useful discussions and insights into the data interpretation and
fractionation processes and equipment. He is the author of three books,
analysis. Many of their ideas are reflected in this article.
the distillation equipment chapter in Perrys Handbook, and more than 80
articles, and has taught the IChemE-sponsored Practical Distillation
This article is based on the paper, Impact of Tray Design on Overall Tray
Technology course more than 300 times. A recipient of several awards,
Efficiency, by H. Z. Kizter, presented at the Distillation 2008 Topical
Kister obtained his BE and ME degrees from the Univ. of New South Wales
Conference, AIChE Spring National Meeting, New Orleans, LA,
in Australia. He is a Fellow of IChemE and AIChE, and serves on the FRI
Apr. 610, 2008.
Technical Advisory and Design Practices Committees.