CHERD-1481; No.

of Pages 5
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

journal homepage: www.elsevier.com/locate/cherd

Predicting valve tray efficiency

Anand N. Vennavelli a,∗ , James R. Whiteley b,1 , Michael R. Resetarits a,2

a Fractionation Research, Inc., Stillwater, OK 74074, United States
b Department of Chemical Engineering, Oklahoma State University, Stillwater, OK 74078, United States

a b s t r a c t

Existing literature models for predicting the mass transfer efficiencies of binary hydrocarbon, distillation columns
employing moving valve trays are evaluated. Only four models for predicting valve tray efficiencies exist in the open
literature. All of these models use data from valve trays. The last theoretical model was published in 1972, 42 years
ago. By comparison, sieve tray efficiency models are numerous and recent. Sieve tray models were developed from
large databases. There are no valve tray equivalents to the fundamental mechanistic models available for sieve trays.
Despite the differences between valve and sieve trays, many of the phenomena on sieve and valve trays are similar.
Consequently, sieve tray models can be employed to provide estimates for valve trays. In this work, using public FRI
data on round moving valves, the performance of the Chen and Chuang sieve tray mechanistic model is compared
to the performance of four valve tray models. It appears that, in the absence of fundamental (and qualified) valve
tray models, the sieve tray models present a potential alternative for valve tray efficiency predictions.
© 2014 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

Keywords: Distillation; Efficiency; Valve trays; Sieve trays; Mass transfer; Modeling

1. Introduction conditions. The empirical models express efficiency either as

Valve trays are the most popular choice for trayed distillation
column internals in the industry (Kister, 1992). Predicting the
mass transfer efficiency or the tray efficiency of valve trays
therefore has significant economic and process design impli-
cations.
Often, the best source of tray efficiency information is expe-
rience with a similar service. However, efficiency data for valve
trays are limited. Vital et al. (1984a,b) compiled a list of lab-
oratory and industrial tray efficiencies including valve tray
efficiencies, but that list had not been updated in 30 years.
When efficiency data are unavailable from an industrial appli-
cation, efficiency is estimated using tray efficiency models.
Even when efficiency data are available, tray efficiency mod-
els can provide valuable information on the effect of design
changes.
Tray efficiency models can be broadly classified as empir-
ical or theoretical. The empirical models are data-driven and
developed to describe the experimental efficiency data in
terms of the physical properties, tray geometry, and operating
a functional relationship or as relationships of dimensionless
groups. The theoretical models are based on phenomenolo-
gical relationships developed from the analysis of the phase
characteristics, mass transfer resistances, and the cross-flow
hydraulic effects on the tray.
The published literature has only four valve tray efficiency
models – the theoretical Todd and Van Winkle (1972) model,
and the empirical Tarat et al. (1974), Scheffe and Weiland (1987)
and Peytavy et al. (1990) models. The only theoretical valve tray
model was published more than 40 years ago.
Most of the tray efficiency modeling work was focused on
sieve trays. The most recent sieve tray efficiency model was
published by Syeda et al. (2007). Due to the similarities in the
operating characteristics of sieve and valve trays, many of the
hydraulic and mass transfer studies on sieve trays are poten-
tially applicable to valve trays. Theoretical sieve tray efficiency
models, which are developed from an improved understand-
ing of the two-phase characteristics and mass transfer on a
tray, and a wide range of data, may provide approximations
for valve trays.

∗
Corresponding author. Tel.: +1 405 385 0354; fax: +1 4053850357.
E-mail addresses: Vennavelli@fri.org (A.N. Vennavelli), rob.Whiteley@Okstate.Edu (J.R. Whiteley), Resetarits@fri.org (M.R. Resetarits).
Received 28 August 2013; Received in revised form 16 December 2013; Accepted 14 January 2014
1
Tel.: +1 405 744 5280.
2
Tel.: +1 405 385 0354.
0263-8762/$ – see front matter © 2014 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cherd.2014.01.022

Please cite this article in press as: Vennavelli, A.N., et al., Predicting valve tray efficiency. Chem. Eng. Res. Des. (2014),
http://dx.doi.org/10.1016/j.cherd.2014.01.022
CHERD-1481; No. of Pages 5
ARTICLE IN PRESS
2 chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx
Physical properties and tray geometries affect tray efficien-
cies. However, the physical properties are the most dominant.
Early empirical efficiency correlations, such as the O’Connell
(1946), were based solely on physical properties. According to
Kister (1992), the O’Connell correlation is still the most popular
efficiency correlation. The dominant effect of physical proper-
ties on tray efficiencies provides another reason to apply sieve
models to valve trays.
The hypothesis of this work is that in the absence of the-
oretical valve tray models, theoretical sieve tray models that
capture the effects of physical properties on tray efficiencies
are sufficiently accurate for valve tray efficiency predictions.
2. Literature survey
A brief review of the four valve models is presented in this
section.
The different databases used to develop the valve models
define the extrapolation-ability and the applicability of the dif-
ferent models. The size, scale, and the nature of the different
databases are indicated in Table 1. The Todd and Tarat models
employed distillation data. The Scheffe and Peytavy models
were developed for absorption systems and from absorption
data.
The only theoretical model for the prediction of valve tray
efficiency was developed by Todd and van Winkle. The data
used for the Todd model were collected from a 0.46 m diameter
column with three rectangular valve trays using benzene/n-
propanol and n-propanol/toluene binaries at atmospheric
pressure. The model was developed using 98 data points.
Three weir heights (25.4 mm, 50.8 mm, 76.2 mm) and three L/V
ratios (0.5, 1.0, 1.5) were studied.
Tarat et al. proposed an empirical model using data col-
lected in a 0.25 m diameter column using the ethanol–water
system at total reflux using Glitsch ballast and annular ballast
trays. The effect of operating conditions, physical properties,
and geometry are considered in terms of dimensionless num-
bers. The data used for the model span a deck open area range
of 7.7–13%, weir heights of 25–70 mm, and liquid rates per bub-
bling area of 1.5–9.0 m3 /m2 /h. The model was developed using
42 data points.
Scheffe and Weiland studied the mass transfer character-
istics of square trays with 0.372 m2 of bubbling area equipped
with standard Glitsch V-1 ballast valves. An operating pressure
of 100 kPa was chosen to ensure that the resistance to mass
transfer was predominantly in the gas phase. The absorption
of SO2 from air into NaOH was used to correlate the volumet-
ric mass transfer coefficient. The model was developed using
140 data points.
Peytavy et al. proposed a model for mass transfer in
gas–liquid absorption systems based on data from a 0.45 m
diameter column using Glitsch V4R valve trays. The data used
for the volumetric mass transfer coefficient were obtained
from the absorption of SO2 from N2 into NaOH. The model
was developed using 72 data points.
Table 1 – Valve data used for the four valve tray efficiency models.
Valve model Column diameter/side (m) Valve type
Todd (1972) 0.46 Unknown
Tarat (1974) 0.25 Glitsch ballast; annular ballast
Scheffe (1987) 0.61 Glitsch V-1
Peytavy (1990) 0.45 Glitsch V4R
Please cite this article in press as: Vennavelli, A.N., et al.,
http://dx.doi.org/10.1016/j.cherd.2014.01.022
The Todd model was based on the two film theory and
incorporated heat effects. The key feature of this model is the
correlation for predicting the heat transfer across the interface
using interfacial and bulk temperatures. However, the inclu-
sion of heat effects increases the complexity of the model.
Lockett (1986) observed that including the heat effects does not
necessarily provide improvements in efficiency predictions for
binary distillation systems. Furthermore, the equations for
predicting heat transfer across the interface are empirical and
cannot be extended to other geometries or systems.
The interfacial area, clear liquid height, and the vapor con-
tact time expressions used in the Todd model are empirical.
Issues and potential implications are described in the follow-
ing paragraphs.
The constants of the Hughmark (1965) interfacial area cor-
relation used in the Todd model were determined from bubble
cap tray data. Fifteen years after the publication of the Todd
model, Scheffe and Weiland (1987) reported interfacial areas
on Glitsch valve trays that were twice the interfacial areas on
bubble cap trays. Consequently, other elements of the Todd
model may be overstated to compensate for the underesti-
mated interfacial areas.
The Todd model used an empirical expression to calcu-
late the vapor contact time. The Todd vapor contact time
correlation is internally inconsistent because Hughmark’s
interfacial area correlation used in the vapor contact time
correlation already assumes that the vapor contact time is
equal to the vapor residence time. Furthermore, the uncer-
tainties in the calculated mass transfer coefficients also affect
the back-calculated vapor contact times and consequently the
correlation.
The clear liquid height correlation used in the Todd model
was regressed from valve tray data. However, Lockett (1986)
compared the performance of the various clear liquid height
models for the air–water system and found that the predic-
tions using the Todd clear liquid height model for the air–water
system do not match well with other valve tray clear liquid
height correlations.
In spite of its limitations, the Todd model is the only funda-
mental valve tray efficiency model available in the literature.
The Scheffe and Peytavy models were developed for pre-
dicting volumetric mass transfer coefficients. The vapor phase
volumetric mass transfer coefficients predicted by these mod-
els must be transformed into efficiencies using a transfer unit
relationship. The method of Huml and Standart (1966) was rec-
ommended by Scheffe under the conditions of complete liquid
mixing and negligible liquid phase resistance. The limitations
of the transformation method introduce an additional degree
of uncertainty in the model predictions.
Both the Scheffe and the Peytavy models focus on vari-
ables representing operating conditions and not on physical
properties and valve tray geometries. This may be appro-
priate for absorption, but it limits the potential applicability
to distillation. In general, the efficiencies in absorption are
lower than those in distillation, and therefore, the extension
of absorption-based correlations to distillation may not be
Binary system Database size
Benzene/n-propanol; 98
n-propanol/toluene
Ethanol/water 42
SO2 from air into NAOH 140
SO2 from N2 into NaOH 72
Predicting valve tray efficiency. Chem. Eng. Res. Des. (2014),
CHERD-1481; No. of Pages 5
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx
Table 2 – FRI experimental data – tray details.
Koch Flexitray
Column diameter (m) 1.22
Number of trays 10
Outlet weir height (mm) 76.2
Number of valves 122
Bubbling area (m2 ) 0.86
accurate (Lockett, 1986). Models developed from absorption
data were not recommended for distillation systems except
when the liquid phase mass transfer resistance is negligible
(Chen and Chuang, 1995).
Of the four valve models, only the Tarat model incorporates
valve geometries as model parameters using dimensionless
geometric ratios – the ratio of the froth height to the maximum
valve lift and the ratio of the vertical slot area to the horizontal
orifice area. For both ratios, the model predicts greater effi-
ciencies with higher ratios. The effect of these dimensionless
groups can likely be explained as follows: the vapor residence
time for a given superficial gas velocity is proportional to the
ratio of the froth height to the maximum valve lift. An increase
in efficiency with an increase in this ratio is therefore a con-
sequence of increased residence time. The interfacial area is
proportional to the ratio of the slot area to the orifice area
(Liang et al., 2008). An increase in efficiency with an increase
in this ratio is therefore a consequence of increased interfacial
area.
The coefficients used in the Tarat model, however, are
limited by a lack of range of systems and valve geometries
used for model development. Furthermore, models developed
from aqueous system data cannot be reliably extended to pre-
dict hydrocarbon distillation efficiencies.
3. Valve tray efficiency data
The FRI test data used in this study are limited to
the cyclohexane/n-heptane (C6/C7) system at 0.34 bar and
1.65 bar, and the iso-butane/n-butane (IC4/NC4) system at
11.4 bar, and to data collected on round moving valves.
The FRI test data, collected between 1956 and 1959,
were obtained using three round valve tray designs: Koch
Flexitray (Fractionation Research, 1956), Glitsch ballast tray
(Fractionation Research, 1958), and Glitsch ballast tray V-1
(Fractionation Research, 1959). A total of 65 points were used
after points with obvious weeping and entrainment were
excluded. The tray details are shown in Table 2.
The range of vapor loads of the FRI experimental data used
in this study are listed in Table 3 in terms of the C-factor, which
is defined as:

G
C-factor, Cs = us (1)
L − G
where us is the vapor velocity based on the bubbling area (m/s),
G and L are the vapor and liquid densities (kg/m3 ).
Table 3 – FRI experimental data – C-factor ranges.
System C-factor range (m/s)
C6/C7 0.34 bar 0.013–0.108
C6/C7 1.65 bar 0.014–0.133
IC4/NC4 11.4 bar 0.013–0.105
Please cite this article in press as: Vennavelli, A.N., et al.,
http://dx.doi.org/10.1016/j.cherd.2014.01.022
3
Glitsch ballast tray Glitsch ballast tray V1
1.22 1.22
9 9
76.2 50.8
124 136
0.88 0.86
All of the efficiency data were collected under total reflux
conditions. The FRI data were reported as overall efficiency
data. These data were converted to point efficiency data using
the method of AIChE, as suggested by Chan and Fair. A typical
plot of point efficiency versus C-factor is shown in Fig. 1.
In addition to the four valve tray models, several sieve tray
efficiency models are available in the literature. Among others,
Chan and Fair (1984), Lockett (1986), Kister (1992), and Bennett
et al. (1997) provide reviews of sieve tray efficiency models. For
comparison purposes, the sieve tray model developed by Chen
and Chuang (1993), which was based on the two-film theory
and was developed from FRI distillation data on sieve trays,
was used to predict the efficiency of FRI valve tray tests.
The key equations of the Chen model are as follows:
1/3  ) 0.5
11(1/0.1 )0.14 [(L FS2 / 2 )] (DG tG
NOG = 0.5
(2)
(11/14)(DG G /DL L ) (ML L/MG G) + 1
EOG = 1 − exp(−NOG ) (3)
where EOG is the point efficiency, NOG is the number of over-
all vapor phase transfer units, L and G are liquid and vapor
densities (kg/m3 ), DL and DG are liquid and vapor diffusivi-
ties (m2 /s),  is the liquid viscosity (Pa s),  is the fractional
0.5
open area, FS is the F-factor defined as us (G ) , us is the vapor
velocity based on the bubbling area (m/s), tG  is the vapor con-
tact time (s), ML and MG are the liquid and vapor molecular
weights (kg/kg mol), and L and G are the liquid and vapor
molar flow rates (kg mol/s),  = mG/L where m is the slope of
the vapor–liquid equilibrium line.
4. Results and discussion
Fig. 2 through 5 show the parity plots of the FRI valve tray effi-
ciency data against the model predictions of the Tarat, Scheffe,
and Peytavy valve tray models, and the Chen sieve tray models.
The Todd model required both liquid and vapor bulk composi-
tions on each tray. Vapor compositions were unavailable from
the FRI data. Assumptions of the vapor compositions, from
simulations for example, would invariably already involve
efficiency assumptions. Therefore, the Todd model was not
included in the analysis.
Fig. 2 shows the parity plot of the Tarat model for the three
valve tray designs tested by FRI. The Tarat model underpre-
dicts the efficiencies for all three tray designs. The model
results can be attributed to several factors:
1. The Tarat model is limited to a maximum percent open
area of 13%. However, all of the FRI data are at open areas
greater than 13%.
2. The FRI data do not exhibit a strong open area effect on tray
efficiency as suggested by the Tarat model.
3. The Tarat model is based on ethanol–water data and does
not include any liquid physical properties. The differences
in the properties of aqueous and hydrocarbon systems are
Predicting valve tray efficiency. Chem. Eng. Res. Des. (2014),
CHERD-1481; No. of Pages 5
ARTICLE IN PRESS
4 chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx
Fig. 1 – A typical plot of point efficiency versus C-factor.
Fig. 2 – Parity plot of the Tarat model.
significant, and, as a result, application of this model to
hydrocarbon systems would not be expected to provide
accurate predictions.
4. The model is based on a limited dataset, and due to the
empirical nature of the model, it may not be applicable to
other systems or beyond conditions in the dataset used for
model development.
Fig. 3 shows the parity plot of the Scheffe model for
the three valve tray designs. The Scheffe model overpredicts
the FRI efficiencies for all trays. The model results can be
attributed to the following factors:
1. The model does not account for the physical property
effects accurately. However, this model was not developed
for generalized valve tray efficiency predictions.
Fig. 3 – Parity plot of the Scheffe model.
Please cite this article in press as: Vennavelli, A.N., et al.,
http://dx.doi.org/10.1016/j.cherd.2014.01.022
Fig. 4 – Parity plot of the Peytavy model.
2. The model does not consider liquid phase resistance. The
model was developed for absorption systems where liq-
uid phase resistance is negligible, but not for distillation
systems.
The model was developed for studying the effect of operat-
ing parameters on efficiency and not as generalized valve tray
efficiency prediction model, or, for distillation applications.
Fig. 4 shows the parity plot of the Peytavy model for the
three valve tray designs. The Peytavy model, like the Scheffe
model, consistently overpredicts the FRI valve tray efficiencies.
The explanatory factors discussed for the Scheffe model also
apply for the Peytavy model.
Fig. 5 shows the parity plot of the Chen sieve tray model for
the three valve tray designs tested by FRI. The Chen sieve tray
model provided better predictions for the FRI efficiencies than
any of the valve tray models.
Fig. 5 – Parity plot of the Chen model.
Predicting valve tray efficiency. Chem. Eng. Res. Des. (2014),
CHERD-1481; No. of Pages 5
ARTICLE IN PRESS
chemical engineering research and design x x x ( 2 0 1 4 ) xxx–xxx
Table 4 – Mean absolute relative percentage errors – FRI
data and model predictions.
Model MARE (%)
Tarat 65
Scheffe 34
Peytavy 29
Chen 9
The performance of each model was quantified in terms
of mean absolute relative percentage error (MARE – Eq. (4)).
Results are presented in Table 4. The Chen sieve tray model
predicts the valve tray efficiencies better than the other
models. Although the Chen model does not have any valve
geometry information, it is able to predict the valve tray effi-
ciencies to a mean absolute relative error of 9% – the lowest of
all models.
1   |EOG, calculated − EOG, measured | 
MARE = × 100
n EOG, measured
This result, however, cannot be over-generalized because
the FRI data used in this study consisted only of hydrocar-
bon distillation data. Except for the Chen model, which was
developed from straight hydrocarbon data, the other valve
tray models were developed using data from different sys-
tems. Nevertheless, it appears that the sieve tray efficiency
models may provide a reasonable alternative in the absence
of qualified valve tray efficiency models for the systems of
interest.
The data used to develop the Chen model have similar
physical properties and operating conditions to the FRI valve
tray data. This suggests that sieve tray models developed from
datasets encompassing systems and operating conditions of
interest may provide reasonable approximations of valve tray
efficiencies.
5. Conclusions
Other than the Todd and van Winkle model, the existing
open-literature valve tray efficiency models are empirical,
based on limited data, and do not effectively incorporate valve
geometries and physical property effects. They were, however,
developed for specific purposes and not as generalized mod-
els for valve tray efficiency predictions. Furthermore, models
developed from absorption may be inadequate for predicting
distillation efficiencies.
The effects of physical properties and geometries on valve
tray efficiencies are not adequately captured by any of the lit-
erature valve tray models. Except the Tarat model, none of
the literature valve tray models incorporate valve geometry
parameters.
The Chen and Chuang sieve tray model predicts the FRI
valve tray efficiencies better than the existing valve tray
Please cite this article in press as: Vennavelli, A.N., et al.,
http://dx.doi.org/10.1016/j.cherd.2014.01.022
5
efficiency models. However, the existing valve tray models
were not developed as generalized valve tray efficiency
models or for distillation applications. Results of applying the
models to hydrocarbon test data from FRI indicate that in the
absence of valve tray efficiency models, the sieve tray models
present reasonable alternatives if they are developed from a
wide range of data and operating conditions.
References
Bennett, D.L., Watson, D.N., Wiescinski, M.A., 1997. New
correlation for sieve-tray point efficiency, entrainment, and
section efficiency. AlChE J. 43, 1611–1626.
Chan, H., Fair, J.R., 1984. Prediction of point efficiencies on sieve
trays 1. Binary-systems. Ind. Eng. Chem. Process Des. Dev. 23,
814–819.
Chen, G.X., Chuang, K.T., 1993. Prediction of point efficiency for
sieve trays in distillation. Ind. Eng. Chem. Res. 32, 701–708.
Chen, G.X., Chuang, K.T., 1995. Liquid-phase resistance to
(4) mass-transfer on distillation trays. Ind. Eng. Chem. Res. 34,
3078–3082.
Fractionation Research, Inc., 1956. Topical Report No. 10.
Obtainable from Oklahoma State University Library Special
Collections and Archives, Stillwater, OK.
Fractionation Research, Inc., 1958. Topical Report No. 15.
Obtainable from Oklahoma State University Library Special
Collections and Archives, Stillwater, OK.
Fractionation Research, Inc., 1959. Topical Report No. 19.
Obtainable from Oklahoma State University Library Special
Collections and Archives, Stillwater, OK.
Hughmark, G.A., 1965. Point efficiencies for tray distillations.
Chem. Eng. Prog. 61, 97–100.
Huml, M., Standart, G., 1966. Efficiency of large Turbogrid trays.
Br. Chem. Eng. 11, 1370.
Kister, H.Z., 1992. Distillation Design. McGraw Hill, New York.
Liang, Y.C., Zhou, Z., Shao, M., Geng, J., Wu, Y.T., Zhang, Z.B., 2008.
The impact of valve tray geometry on the interfacial area of
mass transfer. AlChE J. 54, 1470–1477.
Lockett, M.J., 1986. Distillation Tray Fundamentals. Cambridge
University Press, Cambridge.
O’Connell, H.E., 1946. Plate efficiency of fractionating columns
and absorbers. Trans. AIChE 42, 741–755.
Peytavy, J.L., Huor, M.H., Bugarel, R., Laurent, A., 1990. Interfacial
area and gas-side mass-transfer coefficient of a gas–liquid
absorption column-pilot-scale comparison of various tray
types. Chem. Eng. Process. 27, 155–163.
Scheffe, R.D., Weiland, R.H., 1987. Mass-transfer characteristics of
valve trays. Ind. Eng. Chem. Res. 26, 228–236.
Syeda, S.R., Afacan, A., Chuang, K.T., 2007. A fundamental model
for prediction of sieve tray efficiency. Chem. Eng. Res. Des. 85,
269–277.
Tarat, E.Y., Vasin, N.V., Segal, V.A., ZadorskiVm Olemberg, V.I.,
1974. Efficiency of new designs of valve trays in rectification of
binary mixtures. Int. Chem. Eng. 14, 638–640.
Todd, W.G., Van Winkle, M., 1972. Correlation of valve tray
efficiency data. Ind. Eng. Chem. Process Des. Dev. 11, 589–604.
Vital, T.J., Grossel, S.S., Olsen, P.I., 1984a. Estimating separation
efficiency 1. Introduction. Hydrocarbon Process. 63, 55–56.
Vital, T.J., Grossel, S.S., Olsen, P.I., 1984b. Estimating separation
efficiency 2. Plate columns. Hydrocarbon Process. 63, 147–153.
Predicting valve tray efficiency. Chem. Eng. Res. Des. (2014),