Professional Documents
Culture Documents
A. R. Secchi
Chemical Engineering Program, COPPE/UFRJ, Rio de Janeiro, RJ - Brazil
T. F. Mendes
Dept. of Chemical Engineering, UFRGS, Porto Alegre, RS - Brazil
DOI 10.1002/aic.12166
Published online January 20, 2010 in Wiley Online Library (wileyonlinelibrary.com).
The main objective of this work is to establish appropriated ways for estimating the
overall efciencies of industrial distillation columns with valve trays with downcomer
and dualow trays. The knowledge of efciencies has fundamental importance in the
design and performance evaluation of distillation columns. Searching in the literature,
a tree of alternatives was identied to compose the tray efciency model, depending
on the mass transfer models, the liquid distribution and vapor ow models on the tray,
the liquid entrainment model, the multicomponent mixture equilibrium model, the phys-
ical properties models, the height of froth on the tray model and the efciency deni-
tion. In this work, different methods to predict the overall efciency of distillation col-
umns with valve and dualow trays were composed and compared with data from
three commercial distillation columns under different operating conditions. The models
were inserted in the Aspen Plus 12.1 simulator, in Fortran language, together with
tray geometrical data, uid properties and operating data of the distillation columns.
For each column, the best thermodynamic package was chosen by checking the tem-
perature prole and overhead and bottom compositions obtained via simulation
against the corresponding actual data of industrial columns. A modication in the
fraction of holes evaluation that is jetting parameter of the Garcias hydraulic model
of dispersion above the tray was proposed. This modication produced better results
than the original model to predict the fraction of holes that are jetting and in the ef-
ciency of dualow trays and similar results to Garcia model in the efciency evalua-
tion of valve trays. VC 2010 American Institute of Chemical Engineers AIChE J, 56: 2323
2330, 2010
Keywords: distillation columns, tray efciency, hydrocarbon processing
2324 DOI 10.1002/aic Published on behalf of the AIChE September 2010 Vol. 56, No. 9 AIChE Journal
The FJmodied calculation given in Eq. 5 is suggested in
this work to be used in place of the original FJ calculation
of the model of Garcia11/Garcia and Fair.12 It is in accord-
ance with the statement made by Johnson (apud Prado
et al.16) that 60% of the tray holes are under jetting condi-
tion, when the supercial gas velocity based on active area
is at the transition point from froth to spray condition (uAT).
The correlation for the uAT calculation was presented by
Johnson (apud Prado et al.16) and is given in Eq. 6.
The original correlation suggested by Prado and Fair13 for
the calculation of FJ, which was used by Garcia11/Garcia
and Fair,12 is shown in Eq. 8.
uA uA;0
FJ 0 \ FJ \ 1 (8)
uA;100 uA;0
AIChE Journal September 2010 Vol. 56, No. 9 Published on behalf of the AIChE DOI 10.1002/aic 2325
Table 1. Columns Specications
(A) C6 Column (B) Butene-1 Column (C) Propylene Column
Number of trays 60 138 224
Feed tray 30 100 140
Trays spacing, TS (m) 0.5 0.4 0.4
Light key component Benzene butene-1 propylene
Heavy key component Toluene n-Butane propane
Equation of state/Activity NRTL SRK RK-SOAVE
coefcient model
Data base to binary Aspen VLE LIT e Aspen Ethylene Aspen Pure 12
interaction parameters GMEHLING et al.
Feed ow rate (kg/s) 6.86 4.40 6.65
Vapor product ow rate (kg/s) 0.056
Distillate ow rate (kg/s) 3.96 1.46 6.34
Bottom product ow rate (kg/s) 2.91 2.88 0.31
Reux rate (kg/s) 4.22 32.60 116.57
Reboiler heat duty (GJ/s) 0.208 0.700 2.443
Top pressure (MPa) 0.029 0.664 1.147
Bottom pressure (MPa) 0.062 0.785 1.255
Feed temperature ( C) 127.0 35.0 33.1
Compounds Nitrogen, cycle-pentane, Hydrogen, methanol, propane, Ethane, ethylene, propane, propylene,
n-hexane, methyl-cyclepentane, n-butane, 1-butene, cis-2-butene, cycle-propane, propadiene,
benzene, n-heptane, Trans-2-butene, (1,3)-butadiene, methylacetylene, isobutane,
methyl-cycle-hexane, toluene, isobutane, isobutene, cyclebutane, Green-oil hexaneC12
n-Propyl-cycle-pentane, o-xylene, isopentane, MTBE, DME, water
m-xylene p-xyleno, ethyl-benzene,
ethyl-heptane, m-ethyltoluene,
(1,3)-dimethyl-4-ethylbenzene,
n-undecane, water
Composition (%) 0.0125, 0.0500, 2.0000, 0.0700, 0.0030, 0.0300, 0.0018, 0.00001, 4.6410,
6.0200, 48.6467, 0.6200, 11.1000, 41.4963, 10.8700, 95.2754, 0.0570, 0.0150,
2.2700, 22.7000, 0.8500, 33.5000, 0.0017, 2.3500, 0.0065, 0.00035, 0.0029
1.5300, 3.1000, 1.0500, 0.1100, 0.2900, 0.0510,
9.0000, 0.4700, 1.0100, 0.0080, 0.0850, 0.0350
0.3500, 0.3100, 0.0033
Composition base wt mol mol
component efciencies are usually all different. The treat- mined, which is expected to be low, or even nonexistent, in
ment of such a mixture as a pseudo-binary mixture based on the case of valve trays.
two key components is by far the most common procedure
used in practice, according to Lockett,17 who describes the
pseudo-binary approach used here to calculate the slope of Methodology
equilibrium curve (m), where the compositions of the light
and heavy key components in the liquid and vapor phases The following three commercial distillation columns were
were taken into consideration. The slope of equilibrium used as reference in this study and have the specications
curve is used to calculate k. given in Table 1:
(A) C6 Fractionator: its function is to separate a C5C9
hydrocarbon mixture, in a C6 cut at the top (rich in benzene)
Liquid entrainment and weeping and a C7 cut at the bottom.
The liquid entrainment in vapor reduces the tray efciency (B) Butene-1 Fractionator: its function is to separate a C4
because it represents an internal recirculation of the liquid. hydrocarbon mixture, producing butene-1 at the top with a
For dualow trays, the effect of entrainment on efciency minimal amount of n-butane.
was considered using the relationship suggested by Garcia (C) Propylene Fractionator: its function is to separate high
and Fair2 as function of the ooding factor (FF) and purity propylene at the top from propane.
tray capacity (CSB), whereas the Colburns equation (apud In the case of the C6 Fractionator, several activity coef-
Lockett17) together with the Zuiderweg8 correlation were cient models with different binary parameters were tested to
used for valve trays. compare the actual temperature prole with those obtained
In relation to the effect of weeping, the relationship sug- in the simulator and the expected Murphree efciency. It
gested by Garcia and Fair2 was used for dualow trays, was chosen the NRTL model, which describes the equilib-
whereas for valve trays the effect of weeping on the tray ef- rium liquidvapor and liquidliquid solutions of strongly
ciency was not considered because [] there is not an nonideal. This model requires binary parameters that were
equation that is conveniently simple and analogous to the obtained from literature (Gmehling et al.18) and from linear
one suggested by Colburn (1936), designed to determine the regression of experimental data, already included in the
apparent efciency under weeping conditions (Lockett,17 database of the simulator Aspen Plus 12.1 (called VLE
p. 175). In this work, only the weeping ow rate was deter- LIT).
2326 DOI 10.1002/aic Published on behalf of the AIChE September 2010 Vol. 56, No. 9 AIChE Journal
Figure 2. Algorithm implemented in the Aspen Plus 12.1 simulator to calculate the apparent Murphree vapor phase
tray efciency (EaMV or EW1).
Regarding the Butene-1 Fractionator, the best thermody- Butene-1 (dualow tray) and Propylene fractionators (dual-
namic package was chosen by checking the temperature pro- ow tray), and the overall efciency of the corresponding
le and overhead and bottom compositions obtained via sim- industrial columns, obtained by adjustment to plant data are
ulation against the corresponding actual data of commercial given in Figures 35, respectively. For the valve tray col-
scale columns, checking the Murphree efciency expected umn, Figure 3, the prediction of overall column efciency
for this type of tray (dualow), comparing the data of liq- was done using the Garcia11 model, the modied Garcia
uidvapor equilibrium obtained from the simulation with the model (using the FJmodied as proposed in this work), and the
work of Carmichael et al.,19 and following the recommenda- Chan and Fair7 model developed for crossow tray. For the
tions of the technical support of the Aspen simulator 12.1. It dualow columns of Figures 4 and 5, the prediction of over-
was chosen the equation of state of Soave-Redlich-Kwong all column efciency was done using the same models (but
(SRK option in the simulator Aspen 12.1, with the database with the appropriated corrections for liquid entrainment and
Ethylene for the parameters of binary interaction). weeping), except the last one, which was replaced by the Xu
In the case of the Propylene Fractionator, the choice of et al.1 model, developed for counter-current ow tray.
the equation of state was based on the proles of tempera- In all these gures, the effect of the proposed modication
ture and Murphree efciency expected for this type of tray in the calculation of the jetting volume fraction (FJ) can be
(dualow). The results obtained with the equation of state of
Soave-Redlich-Kwong option RK-SOAVE - Pure 12 were
those that best agreed with the aforementioned plant data. In
the Table 1, it is shown a summary of thermodynamic
choices.
The Aspen Plus 12.1 simulator was used for carrying out
the simulations. The implementation of the efciency mod-
els, correlations for froth height on the tray, correlations for
liquid entrainment and weeping, and mixture models of liq-
uid on the tray was done in Fortran 77 programing language
in the calculator block, and the algorithm used is summar-
ized in Figure 2. More details and the full program can be
found in Domingues.20
AIChE Journal September 2010 Vol. 56, No. 9 Published on behalf of the AIChE DOI 10.1002/aic 2327
Figure 4. Comparison of adjusted and predicted overall
efciency of the butene-1 column (dualow
tray).
Figure 5. Comparison of adjusted and predicted overall Figure 7. Parity plot, observed in plant versus eval-
efciency of the propylene column (dualow uated overall efciency calculated using Gar-
tray). cia11 model.
2328 DOI 10.1002/aic Published on behalf of the AIChE September 2010 Vol. 56, No. 9 AIChE Journal
tray efciency (EMV) that can be used to estimate efcien-
cies in new similar distillation column designs in industry.
Notation
AA active or tray bubbling area (AT 2AD), m2
AH hole area of the tray, m2
AJ fraction of small bubbles present in the bulk froth zone
a0 , a Geometrical interfacial area per volume of vapor and
liquid, m2 m3
CSB tray capacity, m s1
De Eddy diffusivity in the liquid phase, m2 s1
d1 jet diameter, m
dBL, dBS arithmetic mean bubble diameter of large and small
bubbles in the zones 2 e 4, m
dBLS Sauter mean bubble diameter of large bubbles in the
zone 3, m
dBSS Sauter mean bubble diameter of small bubbles in the
zone 5, m
dC column diameter, m
Figure 8. Parity plot, observed in plant versus eval- dH hole diameter, m
uated overall efciency calculated using EOC overall column efciency
EOG point efciency
Chan and Fair7 and Xu et al.1 models. EMV Murphree vapor phase tray efciency
EaMV apparent Murphree vapor phase tray efciency,
accounting the effects of entrainment and weeping of
Conclusions liquid in the valve trays
EW1 apparent Murphree vapor phase tray efciency,
The modied Garcia model for estimating the efciency accounting the effects of entrainment and weeping of
by contribution zones of mass transfer proposed in this work liquid in the dualow trays
coupled with the modications suggested by Garcia e Fair2 FF ooding factor
for dualow trays have successfully represented the dualow FJ fraction of active holes that are jetting
FLB fraction of active holes that are issuing large bubbles
tray column performance (counter-current ow). This is one FS supercial F factor based on AA (uAq0:5 0.5
m0.5
G ), kg
of the rst attempts to apply the Garcia11/Garcia and Fair12 s1
model with the modications suggested by Garcia and Fair2 FSB fraction of active holes that are issuing small bubbles
to dualow tray columns. The modication on the FJ param- h1 jet height, m
hCL clear liquid height, m
eter calculation seemed to be coherent with the jetting frac- hF froth height, m
tion expected for such columns. Furthermore, this is also one hW weir height or wave height in dualow trays, m
of the rst attempts for using the correlations suggested by k0 G, k0 L vapour and liquid phase mass-transfer coefcient, m s1
Garcia and Fair2 for estimating the tray capacity, liquid LW weir length, m
entrainment, and weeping in dualow tray columns. MG, ML vapor and liquid mass ow rate, kg s1
MWG, MWL molecular weight of vapor and liquid mixture,
The performance of butene-1 and propylene distillation kg kmol1
columns were better represented by the modied model sug- m slope of equilibrium curve, dy/dx
gested in this work than by the previous models given by NG number of vapor phase mass-transfer units: NG1 (Zone
Garcia11/Garcia and Fair12 and Xu et al.1 1); NG2L (Zone 2, large bubbles); NG2S (Zone 2, small
bubbles); NG3 (Zone 3); NG4L (Zone 4, large bubbles);
It should also be mentioned that for dualow tray col- NG4S (Zone 4, small bubbles); NG5 (Zone 5); NG6 (Zone
umns, the modied Garcia model represented the efciency 6); NGFJ (Zones 1, 2 e 6); NGFLB (Zones 3, 4 e 6);
peak closer to the overall column efciency obtained from NGFSB (Zones 5 e 6)
plant data adjustment than using the original FJ suggested by N0 L number of liquid phase mass-transfer units
Prado and Fair13 and used by Garcia.11 The peak efciency QG, QL vapor and liquid volumetric ow rate, m3 s1
Ts tray spacing, m
is in the ooding factor range from 0.9 to 1.0 in which the tG mean residence time of gas in dispersion, s
studied dualow columns runs during almost 99% of time, tL mean residence time in the liquid phase, s
so it is the most important operating range for the prediction uA supercial gas velocity based on AA or Ab, m s1
of the overall column efciency. uAT supercial gas velocity based on active area at transition
point from froth-to-spray, m s1
The use of the modied Garcia11 model for the C6 col- uA,0, uA,100 supercial gas velocity, based on active area at 0% and
umn (valve tray) still deserves a certain reserve, as it overes- 100% jetting, m s1
timate the FJ and the original Garcia11 model underestimate Pe Peclet number, dimensionless:
this factor. Both models were better than Chan and Fair7 Pe MW ML
dC L W hCL
qL
MWL De
L
model. The satisfactory results in efciency prediction of the
Greek letters
new model developed for crossow tray by Syeda et al.15
suggest its inclusion in further studies regarding the C6 col- e, a vapor and liquid holdup fraction
k ratio of slope
of equilibrium
umn. curve to slope of operating
Another important contribution of this work was the line: k m MW MG
G
ML
MWL
implementation, in the Aspen Plus 12.1 simulator, of the lG, lL Vapor and liquid viscosity, Pa s kg m1 s1 N s
algorithm to calculate the apparent Murphree vapor phase m2
AIChE Journal September 2010 Vol. 56, No. 9 Published on behalf of the AIChE DOI 10.1002/aic 2329
qG, qL vapor and liquid density, kg m3 11. Garcia JA. Fundamental model for the prediction of distillation sieve
r surface tension, N m1 tray efciency: hydrocarbon and aqueous systems. D. Sc. Thesis in
/ fractional free area (AH/AA) Chemical Engineering. Austin: The University of Texas at Austin,
q
1999.
g constant of Eq. 12: g Pe
2 1 4kE Pe 1
OG
12. Garcia JA, Fair JR. A fundamental model for the prediction of dis-
tillation sieve tray efciency. 1. Database Development and 2 Model
Development and Validation. Ind Eng Chem Res. 2000;39:1809
1825.
Literature Cited 13. Prado M, Fair JR. Fundamental model for the prediction of sieve
1. Xu ZP, Afacan A, Chuang KT. Efciency of dualow trays in distil- tray efciency. Ind Eng Chem Res. 1990;29:10311042.
lation. Can J Chem Eng. 1994;72:607613. 14. Bennett DL, Watson DN, Wiescinski MA. New correlation for
2. Garcia JA, Fair JR. Distillation sieve trays without downcomers: sieve-tray point efciency, Entrainment and Section Efciency.
prediction of performance characteristics. Ind Eng Chem Res. AIChE J. 1997;43:16111626.
2002;41:16321640. 15. Syeda SR, Afacan A, Chuang KT. A Fundamental model for pre-
3. OConnell HE. Plate efciency of fractionating columns and absorb- diction of sieve tray efciency. Chem Eng Res Des. 2007;85:269
ers. Trans Am Inst Chem Eng. 1946;42:741. 277.
4. Drickamer HG, Bradford JR. Overall plate efciency of commercial 16. Prado M, Johnson KL, Fair JR. Bubble-to-spray transition on sieve
hydrocarbon fractionating column as a function of viscosity. Trans trays. Chem Eng Prog. 1987;83:3240.
Am Inst Chem Eng. 1943;39:319. 17. Lockett MJ. Distillation Tray Fundamentals. Cambridge: Cambridge
5. MacFarland SA, Sigmund PM, VanWinkle, M. Predicting distillation University Press, 1986.
efciency. Hydrocarbon Process 1972;51:111114. 18. Gmehling J, Onken U, Kolbe B. Vapor-Liquid Equilibrium Data
6. AIChE. Bubble Tray Design Manual. New York: AIChE, 1958. Collection (Aliphatic Hydrocarbons, Aromatic Hydrocarbons).
7. Chan H, Fair JR. Prediction of point efciencies on sieve trays. 1. Frankfurt: Dechema, 1983.
Binary Systems and 2 Multicomponent Systems. Ind Eng Chem Pro- 19. Carmichael LT, Hwang KC, Berry VM, Sage BH. Phase behavior
cess Des Dev 1984;23:814827. in a six-component hydrocarbon system. J Chem Eng. 1973;7:331
8. Zuiderweg FJ. Sieve trays: a view on the state of the art. Chem Eng 336.
Sci. 1982;37:14411464. 20. Domingues TL. Avaliacao da eciencia de pratos valvulados com
9. Chen GX, Chuang KT. Determining the number of gas-phase and downcomer e pratos dualow em colunas de destilacao industriais,
liquid-phase transfer units from point efciencies in distillation. Ind M. Sc. Thesis in Chemical Engineering (in Portuguese), Porto Ale-
Eng Chem Res. 1994;33:907913. gre: Universidade Federal do Rio Grande do Sul, 2005.
10. Klemola KT, Ilme JK. Distillation efciencies of an industrial-scale
i-Butane/n-Butane Fractionator. Ind Eng Chem Res. 1996;35:4579 Manuscript received Feb. 18, 2009, revision received Aug. 24, 2009, and nal
4586. revision received Dec. 10, 2009.
2330 DOI 10.1002/aic Published on behalf of the AIChE September 2010 Vol. 56, No. 9 AIChE Journal