1

CFD SIMULATION OF MULTIPHASE FLOW IN

A SIEVE TRAY OF A DISTILLATION COLUMN

1
J. G. Teleken;
1
L. O. Werle*;
1
C. Marangoni;
1
M. B. Quadri;
1
R. A. F. Machado

1
Federal University of Santa Catarina- Department of Chemical Engineering

* To whom all correspondence should be addressed
Address: Federal University of Santa Catarina- Department of Chemical Engineering -
Laboratory of Control of Processes - Technological Center - University Campus - Trindade.
P.O. Box 476. ZIP Code: 88010-970. Florianpolis, SC. Brazil.
Telephone / fax numbers: +55 48 3721 9554
E-mail: leandro@enq.ufsc.br



Abstract. Many studies involving distillation columns are based on macroscopic models of
conservation of mass and energy. The present work deals with complex hydrodynamics of
sieve trays in pilot plant distillation columns based on control distributed system with action
on heating at intermediaries points, through electrical resistance heat arranged on the surface
of sieve tray (Marangoni and Machado, 2007; Werle, 2007), using fluid dynamics
computational. In this context, the objective of this work was evaluate the influence of the
electrical resistance heat arranged on the surfaces of the sieve trays have taken on
hydrodynamics. It was implemented one three-dimensional mathematical homogeneous
model, biphasic in the commercial code of fluid dynamics computational (CFD), CFX 11
(AEA Technology) for numerical experimentation study. In results can be observed the
influence of electrical resistance heat of surfaces sieve trays on patterns flow, but this does
not harm to hydrodynamics as a whole, therefore helped to mixture and homogenize the
region with active bubbles, by this way the use of electrical resistance heat could be applied
to a control distributed system approach.

Keywords: distillation columns, fluid dynamics model, CFX.

1. I ntroduction

The Petroleum industry is one of the greatest propulsion of the technological advances in
world. This is due the high competitiveness degree existing around of the Petroleum, creating
an exhaustive search for new technologies that enable greater efficiency in their processes.
The process most widely used in this sector is the distillation, and, in this context, the
microscopic knowledge of the mechanism that occurs is very important. Embedded in this
globalized and competitive context, is the development of sophisticated chemical projects and
processes, applying mathematical models that enable better description of their real behavior,
thats why, a greater understanding, in a microscopic level, of the mechanisms that occur in
distillation very relevant, especially referring to the sophistication of the equipment and
processes (Li et al., 2009).
The valve trays are widely used as phase-contacting devices in distillation columns. The
description of the hydrodynamics of valve trays is of great importance in industrial practice.
Based on the hydrodynamics of valve trays, the separation efficiency and overall tray
2

performance will be predicted, for a given set of operating conditions, tray geometry and
system properties. Hydrodynamics of valve trays was reported by many open literatures
(Kister, 1992; Szulczewska et al., 2003). In general, published literature correlations for tray
hydrodynamics were largely empirical.
In recent years, there are considerable academic and industrial interests in the use of
computational fluid dynamics (CFD) to model two-phase flows in some chemical
engineering equipments. The volume-of-fluid (VOF) technique can be used for a prior
determination of multiphase flow on structure packing. Szulczewska et al. (2003), simulated
gasliquid counter-current flow in a plate-type structured packing. Gu et al. (2004)
developed a two-phase flow CFD model using the VOF method to predict the hydrodynamics
of falling film flow on structured packing. Ataki and Bart (2006), simulated the liquid flow
structure on a structured packing element and liquid redistribution at the node of structure
packing with the VOF model. There are many attempts so far to simulate sieve tray
hydrodynamics using CFD. Yu et al. (1998) and Liu et al. (2000) ignored the variations in
the direction of gas flow along the height of the dispersion to simulate the two-phase flow
behavior, and only the hydrodynamics of the liquid flow was obtained. The inter-phase
momentum exchange (drag) coefficient was required to model the hydrodynamics of
multiphase flow on sieve tray. Fischer and Quarini (1998) attempted to describe the three-
dimensional transient gasliquid hydrodynamics, by assuming a constant drag coefficient of
0.44, which was appropriate for uniform bubbly flow. This drag coefficient was not
appropriate for description of the hydrodynamics of sieve trays operating in either the froth or
spray regimes. Krishna et al. (1999) and van Batten and Krishna (2000) described the
hydrodynamics of sieve trays by estimating a new drag coefficient correlation for a swarm of
large bubbles on the basis of the correlation of Bennett et al. (1983) for the liquid hold-up.
Because this correlation over-predicted the liquid hold-up fraction in froth regime, Gesit et
al. (2003) used the liquid hold-up correlation of Colwell (1979), which worked well in the
froth regime, to predict the flow patterns and hydraulics of a commercial-scale sieve tray
(CFD simulation of hydrodynamics of valve tray). There are several studies in the literature
using fluid dynamics modeling for distillation columns, among them, it is worth to emphasize
the researches by Li et al. (2009), Nikou and Ehsani, (2008), Noriler et al. (2007), which,
in most of the cases, studied the behavior of the speed vectors and the distribution of kinetic
energy of liquid-gas flow in sieve trays. The model was solved using the finite volume
method with variables located in a system of coordinates generalized (Maliska, 2004).
At present, CFD is becoming a powerful research and design tool in chemical
engineering. But there are not any reports about simulating the hydrodynamics of valve trays
by using CFD. The major difficulty for this is that valve floats as valve hole gas velocity
changing. When the velocity of gas passing through the valve hole is larger than a certain
value-the critical valve hole velocity, the valve is full open. It means that valve lifts with the
maximum extent and it does not float. In this condition, CFD can be used to simulate the
hydrodynamics of valve trays. In this paper, a three-dimensional transient CFD model was
developed within the two-phase Euler framework for hydrodynamics of a cylindrical full
open valve tray. Then, simulations were carried out with varying superficial gas velocity,
liquid weir loads and weir heights when valves are full open. The objective is proposed the
development, implementation and application of a microscopic model for momentum
conservation subjected to turbulent flow of the vapour phase, to represent fluid dynamics of
vapor-liquid flow in a perforated plate distillation pilot plant, with validation through
correlation of experimental Bennett et. al. (1983), and then examine the influence that the
resistance used to control heating of distributed distillation pilot plant that are on the surfaces
of dishes, exerts on the fluid dynamic flow.

3

2. Mathematical Model

The model considers the flows of gas and liquid in Eulerian-Eulerian framework, where
the phases are treated with transport equations. The equations used for solving this problem
were: Mass Continuity and Momentum Equations. To solve these, it was necessary to add
and use the equation of momentum flow (Liu et al., 2000). It was considered that the
fluctuations (turbulence) consists of small swarms to be formed and to be dispersed, and that
the Reynolds stresses can be linearly related to the mean velocity gradients (eddy viscosity
hypothesis) similar to the relationship between the stress and the strain tensors in laminar
Newtonian flow. It was assumed the model turbulence standard k- turbulence model for
multiphase flow (Nikou and Ehsani, 2008).

2.1. Mass Continuity Equations

( ) ( ) 0 v f f
t
k k k k k
= V +
c
c

(1)

where f
k
,
k
e v
k
, represent volume fraction, macroscopic density and velocity vector for k
phase, respectively. The gas and liquid volume fractions, f
g
and f
l
, are related through the
summation constraint:

1 f f
l g
= +
(2)

2.2. Momentum Equation

( ) ( ) ( ) | | g f M v v f P f v v f v f
t
k k l . g
T
k k k k k k k k k k k k k
+ + V + V V + V = V +
c
c

(3)

where
k
and g represent the molecular viscosity for kth phase and the gravity vector,
respectively. P
k
is the pressure field, having the same value for the gas end for the liquid
phase. The M
g,l
represents the momentum transfer between the gas and the liquid phases and
the additional flux momentum due to the velocity fluctuation, turbulence, was incorporated in
the diffusion term.
For solving Eqs 1-3, it is necessary additional equations relating the interphase
momentum transfer, M
g,l
and additional fluxes for momentum.

( )
l g l g D
g
l g
l , g
v v v v C
d
f
4
3
M

=
(4)

where d
g
is the bubble diameter and CD is the drag coefficient.

2.3. Turbulence Equations
By application of the time average procedure in the Navier Stokes equations, an extra
term appears due to the turbulent fluctuation of the velocities, which needs to be represented
by a constitutive equation. This term is known Reynolds flux, for more equations. We
consider that the fluctuations (turbulence) consists of small swarms to be formed and to be
dispersed, and that the Reynolds stresses can be linearly related to the mean velocity
gradients (eddy viscosity hypothesis) similar to the relationship between the stress and the
strain tensors in laminar Newtonian flow, an effective viscosity can be assumed:

4

t eff
+ = (5)

and the Reynolds fluxes of a scalar are linearly related to the mean scalar gradient.
The standard k- model is connected with the turbulent kinetic energy and its dissipation
rate as following:
|
|
.
|

\
|
=
c


2
k
C
t

(6)

where k is the turbulent kinetic energy and is the dissipation rate of turbulent kinetic
energy. The conservation equation for turbulent kinetic energy and its dissipation rate can be
written as:

( ) ( )
( ) k
k
t
T P r k k U r k r
t
o| o o o o o
o
o o o o o o o
+ c =
|
|
.
|

\
|
|
|
.
|

\
|
V
|
|
.
|

\
|
o

+ - V +
c
c

(7)

(8)
( ) ( )
( ) c
o| o o c o c
o
o
o o
o
o o o o o o o
+ c
c
=
|
|
.
|

\
|
c V
|
|
.
|

\
|
o

+ c - V + c
c
c
T C P C
k
r U r r
t
2 1
k
t


where C1, C2, C, k e 1 are the model constants and
( ) k
T
o|
and
( ) c
o|
T coefficients of
transfer between phases.
A commercial CFD package CFX 11 of AEA Technology, was used to solve the
equations of continuity and momentum for the two-fluid mixture. This package is a finite
volume solver, using body-fitted grids. The used grid is non-staggered. Discretisation of the
equations at the grid is performed using a finite differencing (finite volume) method.
In the Figure 1 two perforated sieve trays of a distillation column are illustrated, which
were studied in the simulations of accomplished CFD.



(a)
(b)


Figure 1 - See lower of the modules with characteristics of the plate (a) without the electric
resistance, (b) with the presence of the electric resistance.

2.4. Boundary and I nitial Conditions
Physical space is being mapped to a cilindrical computational space. boundary conditions
of all frontiers of the physical domain are necessary: at the inlet, uniform profile of velocities
and turbulent properties are imposed, no slip conditions on the wall for both phases and
5

pressure conditions in the outlet were also applied for the two phases. Figure 2 shows the
physical domain used in the simulations.
The Figure 2 shows the configuration of the systems which has been simulated. The
diameter of the tray is 0.21 m with a height of 0.15 m. The length of the weir is 0.16 m,
diameter of resistence is 0.01 m. The liquid enters the tray through a stocking moon opening.
The number of elements in each one of the numeric meshes was of 415164 and 1129497
elements respectively for the system without and with the presence of the electric resistance
on the surface of the plate.



Figure 2 - Physical domain 3-D Multiphase, (a) without the electric resistance, (b) with the
presence of the electric resistance.

Velocity vector equations are being treated as scalar equations, one scalar equation for
each velocity component. All scalar variables are discretised and evaluated at the cell centres.
Velocities required at the cell faces are evaluated by applying an improved RhieChow (Rhie
and Chow, 1983) interpolation algorithm. Transport variables such as diffusion coefficients
and effective viscosities are evaluated and stored at the cell faces. The pressurevelocity
coupling is obtained using the SIMPLEC algorithm (Van Doormal and Raithby, 1984). No
problems with numerical diffusion are anticipated in view of the smallness of the grid and
time steps used. A fully implicit backward differencing scheme was used for the time
integration.
Air, at ambient pressure, and water were used as the gas and liquid phases, respectively.
At the beginning of the simulation, the conditions consist of to fill up liquid until the weir
height, and air up to the weir height at homogeneous temperature equal to T
0
. The velocity
fields and the turbulent properties were also considered as initial conditions for each
simulation will be shown below. The time increment used in the simulations is 0.001 s.
During the simulation the volume fraction of the liquid phase in the gasliquid dispersion in
the system is monitored and quasi-steady state is assumed to prevail if the value of the hold-
up remains constant for a period long enough to determine the time-averaged values of the
various parameters. Typically, steady state is achieved in about 12 s from the start of the
simulations. Simulations have been performed on a Cluster with 10 machines pentium 4 with
processors running at 200 MHz. A typical simulation took about 7 days to simulate 20 s of
tray hydrodynamics.
6

3. Results and Discussion

In this section are presented the results of simulations performed using the two physical
domains studied and analyzed, that is, with and without the presence of electric resistance on
the plates surfaces.
Figures 3a and 3c demonstrate the distribution profile of volumetric fraction of water in
plan XY and in three different heights (1.5, 9, 16.5 cm) in plan XZ respectively without the
presence of the electrical resistances on the surface of plates. Figures 3b and 3d show the
behaviors of velocities of air (in plan XY) and liquid (in plan XZ to 1.5 cm of the surface of
the plate) vectors respectively.



Figure 3 - Distribution of the liquid volumetric fraction of and velocity vectors for
multiphasic flow without the eletric resistence.

7

It can be observed in Figure 3a that the liquid layer retained (hold-up) represented by the
dark blue color (volume fraction of liquid equal to one) suffers a small increase; occurring a
mixture of two phases and foaming formation, because of high gas flow.
The liquid elevation is more pronounced in region near the wall of exit from lower plate
since entrance of air through holes is a control condition with a constant velocity and normal
boundary, which forces the emerge of some road preferences like this one and making a part
of this air arise through the tube of liquid fall.
Figure 3b illustrates the behavior of velocities vectors of gas phase in XY plan. It is
observed that there are regions of large recirculation and turbulence near the walls; also being
possible to see the appearance of some roads preferences with larger velocity of the gas phase
like in the region near the fall barrier of hold-up and also near the wall of liquid fall. This
demonstrates what was verified through observation of the distribution of volume fraction of
liquid in this plan.
Figure 3c confirms the distribution of liquid volume fraction on three different levels in
XZ plan. It is observed that the liquid phase distribution at 1.5 cm below the plate has a less
homogeneous mixture of liquid and vapor phases, with a region of greater concentration of
air close to the barrier of fall of hould-up and near the wall of fall liquid represented by the
color green. That probably happened due to the initial condition and the outline imposed to
the gas entrance flow, which is constant and normal in the border.
When observing the distribution of liquid volume fraction to 1.5 cm of the top plate, there
is greater homogeneity of the mixture, due to the behavior of vapor flow being closer to
reality, that is, in the holes of the top plate it was not imposed an entry air condition because
the air that goes trough the liquid retained in the lower plate is the same that enters in the
holes of the top plate. It is observed in Figure 3d the behavior of velocity vector of liquid
phase at 1.5 cm from the surface of the plate.
It is noticed the existence of some circulation regions near the walls of the plate and also
in the proximities of the tube of liquid fall. This behavior for a distillation plate presents
some operational and efficiency problems but it is worth emphasizing that the study is
focusing in the hydrodynamic of the plate in order to study these phenomena in particular.
To demonstrate the volume fraction distribution of liquid in the active bubbling region (-
0.5 to 0.5 in X[m]), three lines were built on it (one in the center of the plate and two 5 cm
away to the right and to the left of the central line) for both the lower plate and the top plate.


Figure 4 - Distribution volumetric fraction of the liquid surface of the plates in three different
areas on the surface of the plate
8

It is observed in Figure 4 that the distribution of the liquid volume fraction in the lower
plate presents a great variation in the bubbling region, presenting points of minimum fluid
concentration nearer the wall than the hould-up (close to 0.05 m) which is an undesirable
factor for the good operation of the plate, because this way it will occur the presence of
concentration and temperature gradients in the surface of the plate reducing its efficiency. In
the top plate the mixture among the phases happened much more homogeneously and
satisfactory for the good operation of a perforated plate distillation.
The heights of clear liquid values were obtained in a time space safe to ensure the
condition of quasi- stationary-state, determined by the average volumetric fraction of the
liquid times the height of the computational domain. These values are demonstrated in the
Figure 6.9, compared with values obtained by Bennet et. al. correlating experimental (1983).


0
0,005
0,01
0,015
0,02
0,025
0,03
0,2 0,3 0,4 0,5 0,6 0,7 0,8
C
l
e
a
r

L
i
q
u
i
d

h
e
i
g
h
t
,

h
c
l

(
m
)
Superficial gas velocity, us (m/s)
Correlation of Bennett Models CFX

Figure 5 - Comparison among the Model 3-D with the correlation of Bennett et. al. (1983)
for height of clear liquid for a velocity of feeding of water of 0,1 (m/s).

It can be observed in Figure 5 a small difference between values found for clear liquid
height through simulation and the obtained by the experimental correlation of Bennet et. al.
(1983). The 3D homogeneous multiphase model showed a very similar tendency with the
experimental correlation of Bennet., so that the results obtained by the homogeneous model
predict the height of clear liquid in a perforated plate of distillation with an average error of
11,25%.
With the objective of evaluating the influence that electric resistances arranged on the
surface of the plate have in its hydrodynamics one took place new simulations using the
Figure 6 as physical domain to be studied. The results of the simulation performed using the
same air and liquid flows from the previous case as the new physical domain are represented
in the Figure 6.
Figure 6a, showing the distribution of the volume fraction of liquid, illustrates the
formation of a preferential road again for gas passage by all of the liquid drop, however in the
active bubbling area the mixture occurred in a more homogeneous way with a lower elevation
of the hould-up. In Figure 6b, the behavior of the velocity vectors of gas phase is observed, in
the same plan where the profile of liquid volume fraction was demonstrated, showing
circulation zones near the walls. Again it is possible to observe a preferential road to the gas
phase drainage through the liquid fall tube.

9



Figure 6 - Distribuition of the liquid volumetric fraction of and velocity vectors for
multiphasic flow with presence of the electric reisistances.

In Figure 6c, it is observed that to 1.5cm below the surface on the lower plate there is no
homogeneous mixture of phases for the plate with electric resistance because ther are points
with different concentrations of water (or gas), as it can be noticed by the appearance of
regions with higher concentrations of water (strong blue) and regions almost exempt of water
(green colors tending to red). In this situation it does not happen the mass transfer in a
maximized way in bubbling region, and there are gradients of concentration in this region.
However if compared to Figure 3c and 6c, the mixture that occurs in the surface of the plates
with the presence of resistances is more effective than without the presence of these. It is
shown in Figure 6d the behavior of velocity vectors of liquid phase at 1.5 cm of the surface of
the lower plate, where it is possible to observe that circulation regions near the curved walls
of the plate in the region of liquid fall and some small regions near the electric resistance
wall.
10

To demonstrate the distribution of liquid volume fraction in the active bubbling region (-
0.05 to 0.05 in X [m]) for the physical domain with presence of the electric resistance on the
surface of the plate; three lines were built over this area (one in the center of the plate and
two other 5 cm to right and left of center line) for both the lower plate and the upper plate.


Figure 7 - Distribution volumetric fraction of the liquid surface of the plates in three different
areas on the surface of the plate with presence of the electric resistances.

In Figure 7 the distribution of liquid volume fraction in the lower plate shows only a
small region where there is a considerable variation in the volume fraction in liquid phase in
bubbling region near the wall than the hould-up (close to 0,05m). On the top plate the mixture
between the phases occurred in a much more homogeneous and satisfactory way to a good
operation of a perforated plate distillation.
Comparing the results obtained with and without the presence of the electric resistance on
the surface of the plates, it is not observed damage for the distribution of liquid volume
fraction, in other words, the hydronamic flow on the plate is not affected in a noxious way by
the presence of these. It is verified by the analysis of the Figure 4 and 7 that the presence of
the electric resistances even favor the mixture since there is a more homogeneous distribution
of the liquid volume fraction on the plate. From the separation process point of view this fact
is beneficial because it minimizes the possible gradients of concentration and temperature in
this region.

4. Conclusions

It is concluded from the obtained results of this work that the presence of the resistances
on the surface of the plates influences the liquid patterns flow on it, but this influence does
not bring damages to the hydrodynamics, on the contrary, it favors the homogenization of
phases, legitimizing this approach to distributed control so that it is concluded that the use of
electrical resistances on the surface of the plates, used in the control system with distributed
heating. Moreover, it improves the homogenization of mixture of phases in the active
bubbling region, which is very important for maximizing the transfer of mass. From these
results, in subsequent stage a study will be accomplished with mixtures of ethanol and water
in order to analyze the efficiency of the separation process in both domains, apart from
evaluating the effect that the worked electric resistance has on the process of mass transfer in
this unit.

11

Acknowledgments

The authors are grateful for the financial support from the National Agency of the
Petroleum (ANP), and the Studies and Projects Fund (FINEP), by means of the Human
Resources Program of ANP for the Petroleum and Gas sector PRH-34-ANP/MCT.

References

ATAKI, A.; BART, H. J. Experimental and CFD simulation study for the wetting of a
structured packing element with liquids, Chem. Eng. Technol., v.29, p. 336347, 2006.
BENNETT, D. L.; AGRAWAL, R.; COOK, P.J. New pressure drop correlation for sieve tray
distillation columns, AI ChE J ., v. 29, p. 434442, 1983.
COLWELL, C.L. Clear liquid height and forth density on sieve trays, I nd. Eng. Chem. Proc.
Des. Dev, v. 20, p. 298299, 1979.
FISCHER, C.H.; QUARINI, G.L. Three-dimensional heterogeneous modeling of distillation
tray hydraulics, Paper Presented at the AI ChE Annual Meeting, November 1520,
Miami Beach, USA, 1998.
GESIT, G.; NANDAKUMAR, K.; CHUANG, K.T. CFD modeling of flow patterns and
hydraulics of commercial-scale sieve trays, AIChE J , v. 49, p. 910924, 2003.
GU, F.; LIU, C.J.; YUAN, X.G.; YU, G.C. CFD simulation of liquid film flow on inclined
plates, Chem. Eng. Technol,.v. 27, p. 10991104, 2004.
KISTER, H.Z. Distillation Design, McGraw-Hill, New York, 1992.
KRISHNA, R.; vAN BATEN, J.M.; ELLENBERGER, J.; HIGLER, A.P.; TAYLOR, R.
CFD simulations of sieve tray hydrodynamics, Chem. Eng. Res. Des, v. 77, p. 639646,
1999.
LI, X.G.; LIU, D. X.; XU, S. M.; LI, H. CFD simulation of hydrodynamics of valve tray.
Chemical Engineering and Processing: Process I ntensification, v. 48, p. 145-151,
2009.
LIU, C. J. X.; YUAN, G.; YU, K. T.; ZHU, X. J. A fluid-dynamic model for flow pattern on
a distillation tray, Chemical Engineering Science, v. 55, p. 2287-2294, 2000.
MALISKA, C. R. Transferncia de calor e mecnica dos fluidos computacional. Rio de
Janeiro: Livros Tcnicos e Cientficos, 2004.
MARANGONI, C.; MACHADO, R. A. F. Distillation tower with distributed control
strategy: Feed temperature loads. Chemical Engineering an Technoly, v.30, p.1292-
1297, 2007.
NIKOU, M. R. K.; EHSANI, M. R. Turbulence models application on CFD simulation of
hydrodynamics, heat and mass transfer in a structured packing, I nternational
Communications in Heat and Mass Transfer, v. 35, p. 1211-1219, 2008.
NORILER, D.; MEIER, H. F.; BARROS, A. A. C.; WOLF MACIEL, M. R. Thermal fluid
dynamics analysis of gas-liquid flow on a distillation sieve tray, Chem. Eng. J .,
doi:10.1016/j.cej.2007.03.023, 2007.
RHIE, C.M.; CHOW, W.L. Numerical study of the turbulent flow past an airfoil with trailing
edge separation, AI AA J ., v. 21, p. 15251532, 1983.
SZULCZEWSKA,B.; ZBICINSKI, I.; GORAK, A. Liquid flow on structured packing: CFD
simulation and experimental study, Chem. Eng. Technol., v. 26, p. 580584, 2003.
vAN BATEN, J.M.; KRISHNA, R. Modelling sieve tray hydraulics using computational
fluid dynamics, Chem. Eng. J ., v. 77, p. 143151, 2000.

VAN DOORMAL, J.; RAITHBY, G.D. Enhancement of the SIMPLE method for predicting
incompressible flows, Numer. Heat Transfer, v.7, p. 147163, 1984.
12

YU, K.T.; YAN, X.G.; YOU, X.Y.; LIU, F.S.; LIU, C.J. Computational fluiddynamics and
experimental verification of two-phase two-dimensional flow on a sieve column tray,
Paper Presented at the Working Party Meeting on Distillation, Absorption and
Extraction, European Federation of Chemical Engineering, October 57, Cagliari,
1998.
WERLE, L. O. Minimizao dos Transientes atravs do Aquecimento Distribudo em
uma Coluna de Destilao. Dissertao de Mestrado. Programa de Ps-Graduao em
Engenharia Qumica, Universidade Federal de Santa Catarina, 2007(in Portuguese).