You are on page 1of 9

A FUNDAMENTAL MODEL FOR PREDICTION

OF SIEVE TRAY EFFICIENCY

S. R. Syeda, A. Afacan, and K. T. Chuang


Department of Chemical and Materials Engineering, University of Alberta, Edmonton, AB, Canada.

Abstract: A phenomenological model for froth structure is proposed based on the analysis of
froth images of an active sieve tray taken from a 0.153 m distillation column. Froth is defined
as a combination of bubbles and continuous jets that break the surface of froth projecting
liquid splashes and drops above the surface. To estimate the fraction of small bubbles in
froth, a fundamentally sound theoretical expression is derived from turbulent break-up theory.
A new model for predicting point efficiency of cross-flow sieve trays has been developed
based on the hydrodynamics of an operating sieve tray represented by the proposed froth struc-
ture model. This efficiency model is applicable for both froth and spray regime. Fraction of
by-passed or uninterrupted gas jet is considered as the determining factor for froth to spray tran-
sition. The net efficiency is estimated by adding up the contributions of both bubbles and jets
present in the dispersion. The model is tested against the efficiency data of cyclo-hexane/
n-heptane and i-butane/n-butane mixtures.

Keywords: distillation; tray efficiency; froth; turbulent break-up; bubble size

INTRODUCTION the nature of two-phase mixture in the


transition zone and ask for two separate
The simultaneous mass and heat transfer expressions of interfacial area to predict
combined with the complicated two-phase the tray efficiency in these two regimes.
fluid dynamics make distillation formidable to Zuiderweg (1982) and Stichlmair (1978)
conduct any fundamental analysis of distilla- developed their tray efficiency models based
tion. Furthermore, distillation became a well- on this approach.
established industry long before the theory The FRI efficiency data of commercial sieve
of transport phenomena was established. trays, on the other hand, show smooth tran-

Correspondence to: Thus, the common trend of distillation sition of tray efficiency from the weeping to
Dr K.T. Chuang, Department research to date mostly remains empirical, flooding point. This compelled many
of Chemical and Materials semi empirical or mechanistic in nature. researchers to resort to a single efficiency
Engineering, University of Mass transfer efficiency in distillation is model for both froth and spray regimes.
Alberta, Edmonton, AB,
Canada, T6G 2G6.
associated with the fluid dynamics on a Most of the existing tray efficiency models
E-mail: karlt.chuang@ sieve tray that determines the dispersion (AIChE, 1958; Chan and Fair, 1984; Chen
ualberta.ca structure or the contact area between the and Chauang, 1993) are of this type.
gas and liquid phases. The flow regimes on None of the abovementioned models took
DOI: 10.1205/cherd06111
a sieve tray influence the efficiency directly into account the structure of the two-phase
0263–8762/07/ by affecting the interfacial area. Numerous mixture that is generated on the tray in differ-
$30.00 þ 0.00 studies on flow regimes have been done to ent regimes. The only major attempt that con-
understand the hydrodynamic behaviour of siders the dispersion structure in the froth
Chemical Engineering sieve trays. Most of these studies are mainly regime was made by Prado and Fair (1990)
Research and Design
focussed on the transition from froth to spray for the air/water system. They treated the dis-
Trans IChemE, regime. The definition of froth itself is still persion as three regions: a region near the
Part A, February 2007 very vague in the literature. In froth regime, tray where the gas can either be jetting or
the presence of pulsating jets ranges of bubbling, a bulk froth region which contains
# 2007 Institution
bubbles, liquid splashes and droplets give bubbles with bimodal distribution dispersed
of Chemical Engineers
rise to a highly complex dispersion structure. in the liquid and a spray region at the top.
The traditionally perceived picture of the However, they ignored the spray region in
froth regime consists of bubbles in a liquid their detailed mass transfer model. Later,
continuous phase and that of the spray Garcia and Fair (2000a, b) extended this
regime consists of droplets in a gas continu- model to other systems. Their model was
ous phase. These definitions of froth and shown to agree favourably with a wide
spray regime suggest a sudden change in range of data. However, several adjustable

269 Vol 85 (A2) 269–277


270 SYEDA et al.

parameters needed to be introduced at different stages of the


model to emphasize its mechanistic nature. Bennett et al.
(1997), developed a point efficiency model based on the
mechanistic analysis of sieve tray froth height. The model
considers the fluid on the tray to be contained in a liquid-con-
tinuous region near the tray deck and a vapour-continuous
region on top of the liquid-continuous region. The finial simpli-
fied model takes into account the mass transfer of the liquid
continuous region only and thus has limited applicability in
spray regime. A recent study of van Sinderen et al. (2003),
that deals with entrainment and maximum vapour load of
trays presented a two or three layer model of the two-
phase mixture on the tray. This study, although provides a
detailed insight into the dynamics of the froth, particularly
the mechanisms of entrainment formation but was unrelated
to mass transfer efficiency.
From the above discussions it is evident that most existing
correlations for point efficiency are highly empirical and do Figure 2. Froth image of 67 wt% methanol/water mixture on a sieve
not deal with froth dynamics on a sieve tray. The very few tray in a 0.153 m distillation column.
studies, which consider the nature of the dispersion structure
in their models generally, ignore the contribution of drops and
sprays. These models agree with a wide range of data when
using adjustable parameters, but are less applicable in spray
regime on theoretical ground.
In this study, the froth regime is modelled based on the
analysis of froth images taken from a 0.153 m diameter distil-
lation column. The model describes the froth as a combi-
nation of bubbles and continuous jets. At higher gas load,
the jetting fraction dominates and gives rise to the spray
regime. This froth model is further adopted to develop a
fundamental model for predicting sieve tray efficiency. The
efficiency model takes into account the contribution of both
bubbles and jets to the net mass transfer.

MODEL STRUCTURE Figure 3. Schematic representation of froth on an operating sieve


Froth images taken in a 0.153 m diameter distillation column tray.
are shown in Figures 1 and 2. Based on a careful study of
these kinds of froth images, a froth structure has been sche-
matically presented in Figure 3, where froth is shown as a splashes constitute a major part of the froth. A portion of the
combination of jets, bubbles and liquid splashes. The droplets is formed when bubbles break out of the surface of
images (Figures 1 and 2) show that the liquid droplets and the froth. However, the presence of liquid splashes confirms
that some of the gas jets manage to penetrate through the
froth without forming bubbles and generates liquid splashes
at the end of liquid continuous zone. Figure 4 is a more
detailed representation of the froth model, showing both jetting
and bubbling zones. The jetting zone elaborates how some of
the gas jets formed at the sieve tray holes, cross the froth
uninterrupted and throw liquid splashes above by tearing up
the liquid surface. The bubbling zone shows the process of
large and small bubble formation in the froth. Both zones are
present and remain intimately mixed with each other in real
froth. No jetting is achieved at a relatively low liquid flow
rate. This regime is called bubbling regime, which occurs
close to the weeping limit and is of limited significance for com-
mercial sieve tray operation. As the gas load is increased, an
increasingly greater proportion of gas passes the dispersion in
the form of jets. The spray regime occurs when most of the
gas jets formed at the orifice, reach the liquid surface uninter-
rupted and project the liquid up to form small drops. Unlike in
froth regime, where bubbles form a major part of the interfacial
Figure 1. Froth image of pure methanol on a sieve tray in a 0.153 m area, in spray regime drops are the only contributor to the
distillation column. interfacial area. The point efficiency is estimated by combining

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A2): 269– 277
FUNDAMENTAL MODEL FOR PREDICTION OF SIEVE TRAY EFFICIENCY 271

MODEL DEVELOPMENT
In the following sections a method to estimate point effi-
ciency EOG from equations (1) and (2) has been discussed.

Bubbling Zone
Bubbling zone is considered to have bimodal size distri-
bution of bubbles as reported in many studies (Porter et al.,
1967; Ashley and Haselden, 1972; Lockett et al., 1979;
Kaltenbacher, 1982; Hofer, 1983; Klug and Vogelpohl, 1983).
The small bubbles are the secondary bubbles formed by the
turbulent break-up of the primary bubbles originated from the
orifice. The large bubbles are the unbroken primary bubbles
that remain in the froth due to incomplete break-up.
The specific interfacial area, aiG and residence time, tGLB
for the large bubbles in froth can be estimated from the
Figure 4. Froth structure model on an operating sieve tray. following equations, respectively:

6
aiG ¼ (7)
d32L
the contributions from both bubbling and jetting zones that
hf
exist on a tray. tGLB ¼ (8)
ULB
EOG ¼ (1  fi )EB þ fi Ej (1)
Due to complex nature of the process, there are few analytical
where fj is the volume fraction of the gas that bypasses the expressions for any design in distillation literature. The gen-
bubbles as continuous jets, EB and Ej are contributions of bub- eral trend is to use correlations, which are supported by
bling and jetting zone, respectively, to the net point efficiency. reliable experimental data. The following equations are
Due to incomplete break up of the large (primary) bubbles both used to estimate the Sauter mean diameter and raise velocity
large (primary) and small (secondary) bubbles coexist in bub- of the large bubbles formed at the orifice.
bling zone. Thus EB has contributions from both large and
small bubbles, d32L ¼ 0:887D0:846
H u0:21
H (9)
1=6
ULB ¼ 2:5(VLB ) þ ua (10)
EB ¼ (1  FSB)ELB þ FSB  ESB (2)
Where DH and uH are the hole diameter and velocity; VLB is
where FSB is the fraction of small bubbles. the large bubble volume and ua is the gas velocity based
on the tray active (bubbling) area. Equation (9) based on
the bubble size data measured by electronic probes just
above the sieve tray (Prado et al., 1987). Thus the equation
THEORY OF MASS TRANSFER estimates the unbroken primary bubbles in froth. Three differ-
ent liquid systems with nine different tray geometries were
Following expressions can be obtained from two-film
used to generate the bubble size data. This is by far the
theory,
only correlation for primary bubbles on a sieve tray. Equation
NG ¼ aiG kG tG (3) (10) was originally developed for estimating rise velocity of
bubble swarms through a porous bed (Nicklin, 1962). Later
NL ¼ aiL kL tL (4)
Burgess and Calderbank (1975) showed that this equation
adequately predicts rise velocity of large bubbles in froth on
where
sieve trays. This is again the only study done on this topic.
rL Gf The mass transfer coefficient for the liquid phase, kLLB, is
aiL tL ¼ aiG tG (5) modelled with Higbie penetration theory (Higbie, 1935),
rG Lf
 0:5
Here aiG and aiL represent the geometrical interfacial area per DL
kLLB ¼ 1:13 (11)
unit volume of gas and liquid phases, respectively. Assuming tGLB
that the liquid composition does not change vertically and
vapour passes as plug flow without mixing, the overall This is a well-established model used previously by numer-
mass transfer unit can be related to point efficiency as fol- ous studies. The mass transfer coefficient for gas phase,
lows: kGLB, of the large bubbles is estimated from the numerical
solution presented by Zaritzky and Calvelo (Zaritsky and
EOG ¼ 1  exp (  NOG ) (6) Calveio, 1979). This solution was developed for mass trans-
port models in distillation. It was tested against experimental
In the present study, EOG is obtained from the published Mur- data and was applied in efficiency models such as those by
phree efficiency, Emv, data as outlined by Garcia and Fair Prado and Fair (1990) and Garcia and Fair (2000b). The
(2000a). solution is presented as a plot of Peclet number (PeG) of

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A2): 269–277
272 SYEDA et al.

the gas phase versus the asymptotic Sherwood number used this concept for bubble break-up in pipelines. The
(Sh/). Within the range 40 , PeG , 200 the following poly- same approach is applied here for sieve tray analysis,
nomial provides an excellent fit for the experimental data: where a first order bubble breakage rate is assumed. The
breakage rate of large bubbles in froth is given by
Sh/ ¼ 11:878 þ 25:879( log PeG )
dN
 5:64( log PeG )2 (12) ¼ kN (17)
dt

For the range PeG . 200, It was founf that Sh1 had an Here k is the breakage rate constant and N is the number of
essentially constant value of 17.9. large bubbles. Two additional assumptions are made to keep
Froth height, hf, is estimated from Bennett et al.’s (1983) the calculation simple.
correlation for effective froth height,
(1) All large bubbles are bigger than the maximum stable
 0:67 bubble size and are equally susceptible to the break-up
QL
hf ¼ hw þ C (13) process.
W ae
(2) The number of large and small bubbles at any particular
cross section of the froth is constant.
where
2 3 Let us consider that the number of large bubbles entering the
 0:5 !0:91 froth at t ¼ 0 is Ni. Due to turbulent break-up, Ni reduces to Nf
rG
ae ¼ exp412:55 us 5 (14) at t ¼ Dt. Here Dt is the residence time of large bubbles in the
rL  rG
flow field. Therefore, by integrating equation (17) from Ni at
t ¼ 0 to Nf at t ¼ Dt, the following expression is obtained:
and
Nf ¼ Ni ekDt (18)
C ¼ 0:5 þ 0:438 exp (  137:8 hw ) (15)

There are a number of correlations available in literature to Let us consider that the fractions of large and small bubbles
estimate the froth height on a sieve tray. The unique charac- at t ¼ Dt represent the average fraction of large and small
teristic of equation (13), proposed by Bennett et al. (1983), is bubbles in the froth. The number of unbroken large bubbles
that unlike any other correlations it gives effective froth height at t ¼ Dt is given by
i.e., the height of the liquid continuous region. Since in the
present model, froth height is used to estimate the residence Nf ¼ Ni ekDt (19)
time of bubbles in froth, the height of liquid continuous region
calculated by equation (13) gives the appropriate value. For binary breakage,
Other correlations, which give total froth height i.e., the com-
bined height of liquid and vapour continuous region, would Ns ¼ 2(Ni  Nf ) (20)
over estimate the residence time of bubbles.
Using the above information, NGLB and NLLB can be calcu- where Ns is the number of small bubbles formed at t ¼ Dt.
lated from equations (3) and (4). Equation (4) is then used to Thus the volume fraction of small bubbles in froth can be
get the overall mass transfer unit, NOGLB, from which the con- estimated as follows:
tribution of the large bubbles, ELB, to the net efficiency is
obtained by using the equation (6). Ns Vs 2(Ni  Nf )
The portion of small bubbles in froth is considered to reach FSB ¼ ¼ (21)
Ns Vs þ Nf VL 2(Ni  Nf ) þ Nf ðVL =Vs Þ
equilibrium when mass transfer rate is high (Lockett and
Plaka, 1983). Kaltenbacher (1982) also suggested that the Here VS and VL are the volumes of small and large bubbles,
small bubbles get trapped in the froth and leave the froth respectively. Assuming bubbles have spherical shapes, we
practically saturated. In this case, because equilibrium pre- get the following expression for FSB from equations (19)
vails between the vapour and the liquid phase of small and (21):
bubbles, the efficiency of small bubbles becomes unity, i.e.,

ESB ¼ 1 (16) 2(1  ekDt )


FSB ¼ (22)
2(1  ekDt ) þ (d32L =d32S )3 ekDt
In order to estimate the contribution of small bubbles to the
total efficiency, we need to determine the fraction of small The ratio of large bubble diameter to small bubble diameter,
bubbles, FSB, in froth. Due to lack of experimental data d32L/d32S, is obtained from the existing literature. The
and reliable method to estimate this parameter, expression reported diameter ratios are summarized in Table 1.
for FSB has been derived from turbulent break-up theory of From the above table, we find that the most probable value
bubbles. of the ratio d32L/d32S is 5. The breakage rate constant k is a
In any flow field, the FSB is governed by the bubble break- function of the turbulent flow field and the fluid physical prop-
age rate and the bubble residence time in turbulent zone. erties. Hesketh et al. (1991) showed that the measured defor-
Previous theoretical studies (Valentas et al., 1966; Valentas mation times and breakage time of bubbles can be
and Amundson, 1966) dealing with drop size distribution characterized by the natural mode of oscillation of a sphere
assumed that the breakage rate of a drop is of first order with given by Lamb (1932) and proposed the following functional-
respect to the number of drops. Later Hesketh et al. (1991) ity of the rate constant k,

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A2): 269– 277
FUNDAMENTAL MODEL FOR PREDICTION OF SIEVE TRAY EFFICIENCY 273

Table 1. Reported bubble size distribution on an operating sieve tray. small tray by using free trajectory model. However, Raper
Source Small bubble Large bubble Ratio et al. (1979) showed that Fane et al.’s model under-predicts
the tray efficiency when applied for industrial size tray.
Hofer (1983) 5 mm 25 mm 5 Another important attempt to predict mass transfer efficiency
Ashley and Haselden 5 –10 mm 40–80 mm 8
in spray regime was made by Zuiderweg (1982). His semi-
(1972)
Kaltenbacher (1982) 4 mm 25 mm 6 empirical model is based on the FRI experimental data.
Porter et al. (1967) 5 mm 20 mm 4 This is the only model so far that is not case sensitive and
Lockett et al. (1979) 5 mm 25 mm 5 is readily applicable for spray regime. In this study, we have
chosen Zuiderweg’s spray regime model [equations
(29 –34)] to estimate the contribution of jetting zone to the
total mass transfer efficiency in froth regime;
 
3:8 r0:1 0:3 0:6
L rG v 0:13 0:065
k¼ (23) kGj ¼  2 (1 , rG , 80 kg m3 ) (29)
We0:9
cr s0:4 rG rG
Here, v is the rate of energy dissipation in unit mass; Wecr is 2:6  105
kLj ¼ (30)
the critical Weber number given as m0:25
L
 
ahf KOGj
ru2 dmax Ej ¼ 1  exp  (31)
Wecr ¼ (24) us
s  2 0:37
40 F hL FP
ahf ¼ 0:3 bba (32)
where u2 is the mean square velocity of turbulent flow field and F s
dmax is the maximum stable bubble size against turbulent
break-up; r and s are the density and surface tension of the where, Fbba is vapour rate based on active area, F is the ratio
liquid phase, respectively. The values of reported Wecr range of hole area to active area, hL is the clear liquid height and
over an order of magnitude depending on the flow pattern expressed as
responsible for the deformation of the bubble. In distillation, p 0:25
there is no reported value for Wecr. The rate of energy dissipa- hL ¼ 0:6 hW FP (33)
tion, however, is approximately estimated by v ¼ us g (Kawase b
and Moo-Young, 1990); thus the rate constant becomes
and at total reflux
 
3:8 r0:1
L rG
0:3
 0:5
k¼ (us g)0:6 (25) rG
We0:9
cr s 0:4 FP ¼ (34)
rL
The breakage time Dt can be expressed as
The experimental data obtained by Raper et al. (1982) are
Dt ¼ ntGLB (26) used to evaluate the volume fraction of gas that bypasses
the bubbles formation and forms jets, fj and to estimate the
here, n is any value between 0 and 1. Since both n and Wecr net contribution of jetting zone. Following equation is an
are unknowns, we can combine them into single constant: excellent fit for the average value of jetting fraction, fj as a
function of F-factor, Fbba.
n
C00 ¼ (27)
We0:9
cr fj ¼ 0:1786 þ 0:9857(1  e1:43Fbba ) (35)

By multiplying equations (25) and (26) we get


 0:3 
3:8r0:1
L rG
DETERMINING CONSTANT C00
kDt ¼ C 00
(us g)0:6 (28)
s0:4 Constant C00 is determined by comparing the model with
seven sets of FRI data (Sakata and Yanagi, 1979; Yanagi
The constant C00 will be estimated by comparing the model with and Sakata, 1982). These data sets cover two hydrocarbon
the measured efficiency data. systems, cyclo-hexane/n-heptane and i-butane/n-butane,
at five different pressures in two different tray geometries.
The cyclo-hexane/n-heptane system is widely used for test-
Jetting Zone
ing distillation tray performance. The properties of this system
In froth regime, it is difficult to investigate jetting zone are representative of many hydrocarbon systems operated at
separately as jets are intimately mixed with bubbles. No infor- 400 kPa pressure or below. The data sets for this system are
mation is available in literature on the size of jets or droplets taken at two different pressures, 34 kPa and 165 kPa. The
present in froth. In this study, we will treat the jetting zone as data sets for i-butane/n-butane cover three different pressure
spray and use the correlations of spray regime to estimate levels. The measured efficiencies at high pressures
the contribution of jets in froth regime. Although numerous (2068 kPa and 2758 kPa) have been corrected for vapour
studies have been done to determine the onset of spray, entrainment with the down flow liquid (Hock and Zuiderweg,
very few studies have been focussed exclusively on mass 1982). Figure 5 presents the effect of different values of con-
transfer efficiency in this regime. Fane et al. (1977) achieved stant C00 on the estimated point efficiency for the seven sets
some success in predicting efficiency in spray regime on a of FRI data. The average absolute error was calculated by

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A2): 269–277
274 SYEDA et al.

Figure 5. Effect of constant C00 on point efficiency; expressed as Figure 7. Comparison of measured and predicted point efficiencies
average absolute error. for the cyclo-hexane/n-heptane system at 165 kPa (open hole-area
14%).

the following equation,


P
jEstimated  Experimentalj=Experimental
Error% ¼
nnumber of data
(36)

The minimum error was obtained at C00 ¼ 0.16. The reported


theoretical values of Wecr range from 1 to 4.7 (Hinze, 1955;
Lewis and Davidson, 1982). Within this range, Dt varies
from 0.16 tGLB to 0.644 tGLB at C00 ¼ 0.16. The values are
reasonable for obtaining the average bubble size distribution
within the froth.

PREDICTION OF POINT EFFICIENCY


Figure 8. Comparison of measured and predicted point efficiencies for
The present model introduces a new method to estimate the iso-butane/n-butane system at 1138 kPa (open hole-area 14%).
sieve tray efficiency based on a froth structure that describes
the hydrodynamics of an operating sieve tray. The predicted
point efficiencies, EOG, from the proposed model are com-
pared with the FRI data in Figures 6–12. In all cases, predic- model also predicts the smooth transition of EOG from froth to
tions from two earlier models of Chen and Chuang (1993) and spray regime. Under high operating pressures (Figures 10 and
Garcia and Fair (2000b) are also compared with the proposed 11), the breakage rate constant k is high enough to cause
model. The unique characteristics of the new model is that breakage of all large bubbles. This makes the fraction of
unlike the two other models it predicts the trend of efficiency small bubbles FSB unity and gives rise to high point efficiency
change from weeping to flooding point more closely (Figures under such operating condition. The experimentally measured
7–9). The steady decrease in both fraction of small bubbles fraction of bypassed gas is 0.8 at F-factor 2 (Fane et al. 1977).
and bypassed jets results in gradual decrease of the point effi- Beyond this point froth is dominated by spray and the model
ciency, EOG as the F-factor approaches the weeping point. The reduces to Zuiderweg’s model for spray regime. Thus any

Figure 6. Comparison of measured and predicted point efficiencies Figure 9. Comparison of measured and predicted point efficiencies
for the cyclo-hexane/n-heptane system at 34 kPa (open hole-area for the iso-butane/n-butane system at 1138 kPa (open hole-area
14%). 8.3%).

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A2): 269– 277
FUNDAMENTAL MODEL FOR PREDICTION OF SIEVE TRAY EFFICIENCY 275

Figure 10. Comparison of measured and predicted point efficiencies Figure 12. Comparison of measured and predicted point efficiencies
for the iso-butane/n-butane system at 2068 kPa (open hole-area for cyclo-hexane/n-heptane system at 165 kPa (open hole-area
8.3%). 8.3%).

error in predicting EOG beyond F-factor 2 is inherited from


Zuiderweg’s model. The unique feature of the proposed model is that it is
The prediction of Chen and Chuang (1993) model is satis- based on the analysis of tray hydrodynamics (Figures 1–4)
factory for all six sets of data. The interfacial area in this that describes the real situation on a sieve tray. The model
model is estimated from the bubble size distribution. However, includes both bubble and jet contribution to the total point effi-
since the vapour/liquid dispersion in spray regime mostly con- ciency. The often reported bimodal distribution of bubbles in
sists of drops, the model is applicable only to froth regime. froth is explained as the result of incomplete break-up of pri-
The Garcia and Fair (2000b) model predicts the low- mary bubbles in turbulent flow field. The fraction of small
pressure tray efficiency data adequately. However, it predicts bubbles, FSB, is directly estimated by theoretical analysis
significantly lower tray efficiency than the measured values at of the rate of bubble breakage in froth. The only other similar
high pressures. This discrepancy results from the highly effort to estimate FSB was done by Garcia and Fair (2000b).
empirical nature of the model. The model involves a Although their final model agreed with the database favour-
number of equations and at least four adjustable parameters ably, the study failed to identify the source of bimodal
that match about 22 sets of tray efficiency data, mostly bubble size distribution observed in froth, which made their
measured under low or moderate pressures. The under pre- semi-theoretically obtained FSB expression rather arbitrary.
diction of three sets of data out of 22 sets, did not affect the The present model has been developed to incorporate both
final form of the model. Thus, the model is found to be the froth and spray regimes. The fraction of gas that forms
suitable at low and moderate pressures only. continuous jets, fj, is the determining factor of the contribution
Figure 13 compares the overall performance of the three from each of the regimes. For example, in froth regime,
models. The proposed model predicts within +10% for all 0 , fj , 1. As fj increases with higher a gas load, transition
the systems and shows better performance than the two to spray regime occurs gradually and fj becomes unity as
other models. The agreement between the experimental spray regime is reached. No sudden change in dispersion
data and predictions of the new model proves the validity of structure occurs during this transition, and there is a
the proposed approach. smooth transition of FRI efficiency data from froth to spray
regime. Thus the effect of the present approach of consider-
ing the effect of the flow regimes on the tray efficiency
adopted in the proposed model differs from that resulting
DISCUSSION
from the two previous approaches of the existing models.
Tray hydrodynamics is considered to be the key factor in One of the approaches is to apply the same efficiency
determining the nature of two-phase mixture in distillation. model for both froth and spray regimes without considering

Figure 11. Comparison of measured and predicted point efficiencies for Figure 13. Overall comparison of the proposed model with two other
the iso-butane/n-butane system at 2758 kPa (open hole-area 8.3%). existing models.

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A2): 269–277
276 SYEDA et al.

the effect of change of the dispersion structure (AICHE, 1958; dmax maximum stable bubble diameter in turbulent flow
field, m
Chan and Fair, 1984; Chen and Chuang, 1993) The other
d32L sauter mean bubble diameter of large bubbles, m
approach is to use two completely different models for froth d32S sauter mean bubble diameter of small bubbles, m
and spray regime (Zuiderweg, 1982). Since the dispersion DG molecular diffusion coefficient for gas, m2 s21
structure in froth regime is just inverse to that of spray DH orifice diameter, m
regime, applying the same efficiency model for both regimes DL molecular diffusion coefficient for liquid, m2 s21
EB overall point efficiency for bubbling zone
without considering the change in the dispersion structure is Ej overall point efficiency for jetting zone
the incorrect way to estimate the tray efficiency. On the other ELB overall point efficiency for large bubbles
hand, when two separate models are used for the two EOG overall point efficiency (gas composition basis)
regimes difficulties arise in identifying the exact transition ESB overall point efficiency for small bubbles
point. By including the fraction of jetting, dependent on gas fj volume fraction of gas bypasses the froth bubbles as
continuous jet
flow rate, the new model takes into account the difference F ratio of hole to active (bubbling) area
in dispersion structure between the coexisting froth and pffiffiffiffiffiffi
Fbba vapour rate based on active area (ua rG ),
spray regimes. Thus the model provides a logical solution (kg m23)0.5m s21
that can be applied continuously over the range of flow FP flow parameter, (rG =rL )0:5 at total reflux
FSB fraction of small bubbles
rates, without resorting to an arbitrary selection of the use g gravitational constant, 9.8 m s22
of the same or separate models for both the forth and Gf gas mass flow rate, kg s21
spray regimes, and thereby fully describes the smooth hf froth height, m
transition between the regimes. hL clear liquid height, m
The inclusion of physical properties considered in the esti- hW weir height, m
k first order bubble breakage rate constant, s21
mation of fraction of small bubbles FSB [equation (28)] kG gas-phase mass transfer coefficient, m s21
makes the model applicable to systems with wide range of kGj kG for jetting zone
physical properties and under different pressure levels, kGLB kG for large bubbles
where physical properties of the same systems can vary sig- kL liquid-phase mass transfer coefficient, m s21
kLLB kL for large bubbles
nificantly. Moreover, the calculation steps of the proposed kLj kL for jetting zone
model are much simpler and less rigorous than those of KOGj KOG for jetting zone
other similar models (Garcia and Fair, 2000a, b). Lf liquid mass flow rate, kg s21
The present model fully incorporates the jetting fraction of N the number of large bubbles
the dispersion as spray. Due to lack of definitive data on the Ni the number of large bubbles formed at the orifice at any
instant
structure of the spray regime, this study utilized the semi- Nf the number of unbroken large bubbles leaving the froth at
empirical spray regime model of Zuiderweg to estimate the jet- any instant
ting contribution. Thus the current level accuracy of predicting Nf number of unbroken large bubbles remained from Ni at
using the proposed model is limited by the semi-empirical t ¼ Dt
NG number of gas-phase mass-transfer units
nature from Zuiderweg’s model and is not applicable for sys- NGLB NG for large bubbles
tems with vapour density less than unity. More fundamental NL number of liquid-phase mass-transfer units
studies of drop dynamics and quantification of point efficiency NLLB NL for large bubbles
in spray regime will improve the model and enhance the corre- NOG number of overall gas-phase mass-transfer units
NOGLB NOG for large bubbles
lation between the model and experimental data.
Ns number of secondary bubbles formed from Ni at t ¼ Dt
p pitch of holes on sieve plate, m
PeG Peclet number (d32LULB/DG)
CONCLUSIONS QL liquid flow rate, m3 s21
A fundamental model to predict point efficiency has been Sh/ asymptotic Sherwood number (kGLBd32L/DG)
tG mean residence time of gas in dispersion, s
proposed based on the hydrodynamics of an operating tGLB mean residence time of large bubbles in dispersion, s
sieve tray. The new model predicts the FRI efficiency data tL mean residence time of liquid in dispersion, s
of hydrocarbon systems within +10%. It is also able to pre- Dt the time when half of the total secondary bubbles are
dict the trend of tray efficiency from weeping to the flooding formed in froth from the initial Ni number of bubbles, s
point more closely than any other model. The present ua gas velocity based on active(bubbling) area, m s21
uH gas velocity based on total open hole area, m s21
model is based on the analysis of real froth, and so is us gas velocity based on total column cross-sectional
based on sound empirical data, and so the model is more area, m s21
adoptable to the diversified conditions than any other existing ULB rise velocity of large bubbles, m s21
models. u2 mean square velocity of turbulent flow field, (vdmax)2/3, m
The model can be used throughout the froth and spray VLB volume of large bubbles, m3
W weir height, m
regimes and the transition between them, and so will be Wecr critical Weber number
more applicable for the prediction of distillation tray efficiency.
Further fundamental research on point efficiency in spray
regime, however, would make the model more universal.

NOMENCLATURE
Greek symbols
a interfacial area per volume of two-phase mixture, m2 m23 ae froth density defined by equation (14)
aiG geometrical interfacial area per volume of gas, m2 m23 mL liquid viscosity, Pa s21
aiL geometrical interfacial area per volume of liquid, m2 m23 rG gas density, kg m23
b weir length per unit bubbling area, m21 rL liquid density, kg m23
C constant defined by equation (13) s surface tension, N m21

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A2): 269– 277
FUNDAMENTAL MODEL FOR PREDICTION OF SIEVE TRAY EFFICIENCY 277

REFERENCES Lamb, H., 1932, Hydrodynamics, 6th edition (Cambridge University


Press, Cambridge, UK).
Ameican Institute of Chemical Engineers (AIChE), 1958, Bubble Tray Lewis, D.A. and Davidson, J.F., 1982, Bubble splitting in shear flow,
Design Manual (AICHE, New York, USA). Trans I Chem Eng, 60: 283–291.
Ashley, M.J. and Haselden, G.G., 1972, Effectiveness of vapor- Lockett, M.J., Kirkpatrick, R.D. and Uddin, M.S., 1979, Froth regime
liquid contacting on a sieve plate, Trans Inst Chem Eng, 50: point efficiency for gas-film controlled mass transfer on a two-
119– 124. dimensional sieve tray, Trans IChemE, 57: 25– 34.
Bennett, D.L., Agarwal, R. and Cook, P.J., 1983, New pressure drop Lockett, M.J. and Plaka, T., 1983, Effect on non-uniform bubbles in
correlations for sieve tray distillation columns, AIChE J, 29: the froth on the correlation and prediction of point efficiencies,
434– 442. Chem Eng Res Des 61: 119–124.
Bennett, L.D., Watson, N.D. and Wiescinski, M.A., 1997, New corre- Nicklin, D.J., 1962, Two-phase bubble flow, Chem Eng Sci, 17: 693–702.
lation for sieve-tray point efficiency, entrainment, and section effi- Porter, K.E., Davis, B.T. and Wong, P.F.Y., 1967, Mass transfer and
ciency. AIChE J, 43(6): 1611– 1625. bubble sizes in cellular foams and froths, Trans Inst Chem Eng,
Burgess, J.M. and Calderbank, P.H., 1975, The measurement of 45: T265– T273.
bubble parameters in two-phase dispersion-II, Chem Eng Sci, Prado, M., Johnson, K.L. and Fair, J.R., 1987, Bubble-to-spray tran-
30: 1107– 1121. sition on sieve trays, Chem Eng Prog, 83(3): 32– 40.
Chan, H. and Fair, J.R., 1984, Prediction of point efficiencies on sieve Prado, M. and Fair, J.R., 1990, Fundamental model for the prediction
trays, Ind Eng Chem Process Des Dev, 23: 814 –819. of sieve tray efficiency. Ind Eng Chem Res, 29: 1031–1042.
Chen, G.X. and Chuang, K.T., 1993, Prediction of point efficiency for Raper, J.A., Hai, N.T., Pinczewski, W.V. and Fell, C.J.D., 1979, Mass
sieve tray in distillation, Ind Eng Chem Res, 32(4): 701–708. transfer efficiency on simulated industrial sieve trays operating in
Fane, A.G., Lindsey, J.K. and Sawistowski, H., 1977, Operation of a the spray regime Inst Chem Engrs Symp Series No 56, 2.2/57–
sieve plate in the spray regime of column operation, Indian Chem 2.2/74.
Engr, Jan: 45. Raper, J.A., Kearney, M.S., Burgess, J.M. and Fell, C.J.D., 1982, The
Garcia, J.A. and Fair, J.R., 2000a, A fundamental model for the pre- structure of industrial sieve tray froths, Chem Eng Sci, 37(4): 501–
diction of distillation sieve tray efficiency. 1. Database develop- 506.
ment, Ind Eng Chem Res, 39: 1809–1817. Sakata, M. and Yanagi, Y., 1979, Performance of a commercial-scale
Garcia, J.A. and Fair, J.R., 2000b, A fundamental model for the pre- sieve tray, Inst Chem Eng Symp Ser No 56, 3.2/21– 3.2/30.
diction of distillation sieve tray efficiency. 2. Model development Stichlmair, J., 1978, Die Grundlagen des Ga-Flussig-Kontaktappa-
and validation, Ind Eng Chem Res, 39: 1818–1825. rates Bodenkolonne (Verlag Chemie, Weinheim, Germany).
Hesketh, R.P., Etchells, A.W. and Fraser Russell, T.W., 1991, Bubble Valentas, K.J. and Amundson, N.R., 1966, Breakage and coalesc-
breakage in pipeline flow, Chem Eng Sci, 46(1): 1 –9. ence in dispersed phase systems, Ind Engng Chem Fundam, 5:
Hinze, J.O., 1955, Fundamentals of the hydrodynamics mechanism 533– 544.
of splitting in dispersion processes, AIChE J, 1(3): 289–295. Valentas, K.J., Bilous, O. and Amundson, N.R., 1966, Analysis of
Higbie, R., 1935, The rate of absorption of a pure gas into a still liquid breakage in dispersed phase systems, Ind Engng Chem
during short periods of exposure, Trans Am Inst Chem Eng, 31: Fundam, 5(2): 271 –279.
365– 388. van Sinderen, A.H., Wijn, E.F. and Zanting. W.R., 2003, Entrainment
Hoek, P.J. and Zuiderweg, F.J., 1982, Influence of vapour entrain- and maximum vapour flow rate of trays. Trans IChemE, 81: 94–
ment on distillation tray efficiency at high pressure, AIChE J, 106.
28(4): 535–539. Yanagi, Y. and Sakata, M., 1982, Performance of a commercial-scale
Hofer, H., 1983, Influence of gas-phase dispersion on plate column 14% hole area sieve tray, Ind Eng Chem Process Des Dev, 21(4):
efficiency, Ger Chem Eng, 6: 113– 118. 712– 717.
Kaltenbacher, E., 1982, On the effect of the bubble size distribution Zaritzky, N. and Calvelo, A., 1979, Internal mass transfer coefficient
and the gas-phase diffusion on the selectivity of sieve trays, within single bubbles. Theory and experiment, Can J Chem Eng,
Chem Eng Fund, 1(1): 47–68. 57: 58– 64.
Kawase, Y. and Moo-Young, M., 1990, Mathematical models for Zuiderweg, F.J., 1982, Sieve trays: A view on the state of the art,
design of bioreactors: applications of kolmogoroff’s theory of isotro- Chem Eng Sci, 37(10): 1441– 1464.
pic turbulence, The Chem Eng, 43: B19– B36.
Klug, P. and Vogelpohl, A. 1983, Bubble formation with superimposed The manuscript was received 30 March 2005 and accepted for
liquid motion at single-hole plates and sieve plates, Ger Chem publication after revision 1 September 2006.
Eng, 6: 311– 317.

Trans IChemE, Part A, Chemical Engineering Research and Design, 2007, 85(A2): 269–277

You might also like