(423e) Sieve Tray Performance In Heteroazeotropic Distillation for a Viscous System

The objective of this work is to examine
the influence of weir height, and operation conditions on the sieve tray
performance in heterogeneous azeotropic distillation. The process fluids
employed were water, toluene and glycerol.

Glycerol is mainly obtained as a by-product of
biodiesel produced by transesterification. Although, many attempts have been
made to use raw glycerol, in order to decrease the biodiesel production cost,
it is still necessary purify glycerol for majority of applications [1]. Traditional routes of glycerol purification under low
pressures are used to prevent its degradation [2]. The Laboratory of Thermal
and Mechanical Separations in the Chemical Engineering Department has developed
a route to recover glycerol (PI 0804383-3), whose final stage is the azeotropic
distillation. This operation takes advantage of the formation of heterogeneous
azeotrope between water and toluene, allowing dehydration of glycerol at
atmospheric pressure [3].

Weir height is a geometrical variable that increases
the tray efficiency in the froth regime, because it maintains a desired liquid
level on the tray. This level should be high enough to provide sufficient
liquid-vapor contact time and good bubble formation [4]. On the other hand,
vapor and liquid flow rates influence also the tray efficiency, directly, on
the liquid agitation in the tray, or indirectly, through their influence on the
entrainment of liquid. Though, tray efficiency is almost independent of the
vapor flow in a reasonable range of flow rates, when the resistance of the
liquid phase begins to become important this independence disappears [5].
Finally, the significance of the initial concentration of glycerol is related
to the viscosity. An increase in the fluid viscosity decreases the diffusivity
on the liquid phase, leading to a resistance raise in the liquid phase and a
tray efficiency decrease [6, 7].

In this work, experiments were carried out in a column
with three sieve trays and 100 mm of internal diameter (i.d.). The geometrical
specifications of each tray were 3.0% free area, 2.8 mm holes and two circular
weirs with 18 mm of i.d.. Four different weir heights were studied: 10, 25, 50
and 70 mm. During the experiment, an aqueous solution of glycerol was pumped
from the top of the column, and toluene vapor was fed from the bottom of the column.
On the sieve tray were formed two liquid phases: one rich in glycerol disperses
in another liquid phase rich in toluene. The vapor was almost composed by
toluene and water. The reflux liquids were separated into aqueous and organic
phases by decanters at the top and bottom of the column. The organic phase,
toluene, has returned to the column. The water and the final aqueous solution
of glycerol were removed from both the top and the bottom of the column,
respectively, for their flow rates measurement. The final concentration of
glycerol was measure by iodometry [8].

The operational conditions (independent variables)
studied were vapor flow rate, feed flow rate and initial concentration of
glycerol. For each tray configuration, tests were performed according to 2^3-1
fractional factorial design with four center points [9]. The variable response
(dependent variable) was the increment of glycerol content [%wt.]. This is the
result of the difference between final and initial glycerol concentrations. The
levels of independent variables, in original units and coded units, are in
Table 1.

Table 1       Independent
variables for units, coded and original.

Froth regime was visually observed on the tray during
all distillation runs. The statistical analysis was performed using the
Statistica v.9 software. To each analysis were estimated the regression
parameters and the ANOVA for regression. For all tray configurations, the fit
of the regression model was done by determination coefficient (Table 2). The
behaviors of independent variables on the responses have been described for
empirical models; they are shown in equations 1 to 4 in Table 2, valid only
over studied ranges of the independent variables.

Table 2       Models
of increment of glycerol content (Dg) for each weir height (H_w) [mm].

Comparing
the four models in Table 1, it may be observed that both the constant and all
the coefficients increase in absolute values as the weir height increase. The
models also show that X₃ (initial concentration of glycerol coded)
has similar coefficient values for the four configurations. This coefficient is
negative, it means that the increment decreases with increase of initial
concentration of glycerol. For instance, the increment is about 16 percentage
points when the initial concentration is 50%, however the increment is about 5
percentage points when the concentration is 90%. This occurs due to the
dehydration decreases with increase of the viscosity. Besides, the coefficients
of X₁ (vapor flow rate coded) and X₂ (feed flow rate
coded) presented the same order of magnitude in absolute values, but with
contraries signals for each configuration. The best separation of water occurs
to high value of vapor flow rate and low feed flow rate, this behavior is
illustrated in Figure 1.

Fig. 1 Effects
of weir height (H_w) and of operational variables on increment of
glycerol content, in which, X₁ (vapor flow rate coded), X₂
(feed flow rate coded), X₃ (initial concentration of glycerol coded)
are described in Table 1.

The first and second sets of points of the Figure 1
(for low vapor flow rate) show that increasing of dehydration is more
controlled by the feed flow rate, than the initial concentration of glycerol.
In the second set of points, the increase in the increment of glycerol content
is twice, this is the data set in which the weir height exerts major influence.

The third set of points presents the highest values of
increase of glycerol content indicating that they are the best conditions of
the operating variables. In this case, it may be seen also clearly the positive
effect of the weir height as in the previous cases.

In the fourth set of points, where all independent
variables operate to their maximum levels, it is observed that the increase in
glycerol content is relatively low even with high vapor flow rate of the
entrainer. Even so, weir heights from 10 to 70 mm can cause about 40% increase
in the increment.

The difference between the third and the fourth set of
points is the feed flow rate and the initial concentration of glycerol that are
opposite. The major effect between these groups is the initial concentration of
glycerol.

The fifth set of points presents the mean behavior
among all the conditions, in this group only appears the effect of the weir
height.

Comparing the third and the second set of points, it
may be seen clearly that, for these runs, operational conditions were more
significant than weir height.

In conclusion, our results demonstrated that, for
heteroazeotropic distillation with a viscous liquid phase, in spite of having a
positive effect of the weir height, glycerol dehydration depends mainly on the
operational conditions, for the range of 10 to 70 mm of weir height.

References:

[1] F.Skopal, M. Hajek, Treatment of glycerol phase
formed by biodiesel production.  Bioresource
Technol., 101 (2010) 3242 ? 3245.

[2] A. E Bailey, Y. H. Hui, Glycerol processing. In:
Bailey's industrial oil and fat products. 5th ed. N.Y.: John Wiley, 5 (1996) 275-308.

[3] E. Gutiérrez-Oppe, Desidratação por destilação
azeotrópica da glicerina obtida como subproduto da produção do biodiesel. Dissertação
(Mestrado) - EPUSP, São Paulo (2008).

[4] H.Z. Kister.. Distillation design. Cap.6. Tray
Design and Operation. New York : McGraw-Hill..1992

[5] AIChE Research Committee. Bubble Tray Design
Manual. AIChE. New York. 1958.

[6] M. J. Locket. Distillation tray fundamentals. N.
Y. Cambridge (1986).

[7] S. Mahiout, A. Vogelpohl. Mass transfer in high
viscosity media.  Chem. Eng. Process., 18,
4, (1984) 225-232.

[8] IUPAC.
Standard Methods for the Analysis of Oils. Fats and Derivatives Section
III. Glycerol.  Fats and Derivatives 6th
Edition 1st Supplement: Part 23 (1982) Pure & Appl.Chem. 54, 6, (1982)1257?1295..

[9] G. E. P. Box, W. G. Hunter, J. S.  Hunter. Statistic for experimenters: an
introduction to design. data analysis and model building.  2ed.
NY: John Wiley. (2005) 672 p.

[10] H. R. Mortaheb, H. Kosuge, K. Asano, Hydrodynamics
and mass transfer in heterogeneous distillation with sieve tray column, Chem.
Eng.J. 88 (2002) 59-69.