Professional Documents
Culture Documents
GENERAL RESEARCH
Solvent sublation is a nonfoaming wastewater treatment process that combines the benefits of
bubble fractionation and liquid-liquid extraction in a way that does not require mixers, settlers,
or subsequent downstream treatment. A review of past work on small lab-scale batch columns
revealed that removal efficiencies of nonvolatile and volatile organic compounds are generally
higher than those observed in bubble fractionation, air stripping, and conventional liquid-liquid
extraction. In this work, the first of a three-part series, the transport mechanisms in a three-
phase continuous, countercurrent sublation process are presented. Two mathematical models,
namely, the series CSTR model (SCM) and the two-phase axial dispersion model (ADM2), are
developed. It is shown that these two models are equivalent and can be used interchangeably
with the aid of a simple expression. Nondimensional correlations, based upon simulated data
obtained from the SCM, are generated to predict the steady-state fractional removal, FR, and
the separation factor, Σ (ratio of effluent solvent concentration to effluent water concentration),
for strongly hydrophobic compounds. The effects of operational, hydrodynamic, thermodynamic,
and design variables on sublation performance are discussed.
phenols, the use of surfactant can be avoided by adjust- needed, which makes sublation less expensive. The
ing the pH (Valsaraj and Springer, 1986). biggest advantage over both conventional extraction
Solvent sublation may be thought of as a combination (single-staged) and bubble fractionation is that a higher
of conventional liquid-liquid extraction and bubble degree of fractional removal is possible under certain
fractionation. In Figure 1, countercurrent configura- circumstances (Valsaraj et al., 1991).
tions of these processes are compared. Like all adsorp- As can be seen in Table 1a,b, most of the sublation
tive bubble processes, solvent sublation is similar to work to date has been conducted on small-scale labora-
bubble fractionation. The difference is that, in subla- tory columns operating in a batch mode relative to the
tion, a nonvolatile organic solvent is floated upon the liquid phase. Though this type of information is im-
water column. In bubble fractionation, nonvolatile portant for studying the transport mechanisms in
compounds become enriched in the top portion of the sublation, it is not useful for ascertaining scaleup or
aqueous phase, and volatile organic compounds (VOCs) industrial feasibility.
are lost to the atmosphere unless some other gas-phase Recently (Lu et al., 1991; Valsaraj and Thibodeaux,
treatment is used. In solvent sublation, on the other 1991a,b; Valsaraj et al., 1992), our group investigated
hand, volatile and nonvolatile compounds are recovered semicontinuous operation (continuous, countercurrent
in the solvent phase, just like they are recovered in the air and water phases) and examined the effect of column
solvent phase used in the extraction process. In the diameter (15 versus 5 cm). These studies demonstrated
extraction process, multiple stages are required for any the feasibility of operating continuously and showed
appreciable recovery; however, in sublation, only one that solvent sublation consistently outperformed bubble
stage of solvent-water contact is required. Further- fractionation in terms of fractional removal. Further-
more, in solvent sublation, mixers and settlers are not more, sublation performance was found to be less
+ +
Table 2. Thermodynamic Properties of Hydrophobic ematical models to predict the steady-state pollutant
Compoundsa concentrations in all three phases of a continuous
PAH (pyrene) CP (PCP) sublation column. We examine the effects of the
(MW ) 202) (MW ) 266) operating variables, thermodynamic variables, and
aqueous solubility (mg/L) 0.135 14 design variables on sublation performance. Model-
vapor pressure (mmHg) 6.80 × 10-7 1.10 × 10-4 generated, fractional removal data are correlated in
log Kow′′ 4.88 5.01 terms of key dimensionless numbers so that the use of
Henry’s constant, Hc (cm3/cm3) 7.65 × 10-4 1.39 × 10-4 the complex analytical solutions may be avoided. In
Interfacial partition constant, 5.39 × 10-3 9.13 × 10-5 part 2, we present hydrodynamic data, such as bubble
KA (cm)b
size and gas holdup, for two types of gas spargers used
a Montgomery and Welkom, 1990. b Mackay et al., 1990; Hoff
in a pilot-scale sublation column. We also show that
et al., 1993. the homogeneous or bubbly flow regime (Shah et al.,
dependent on the column diameter. This was in con- 1982) is the appropriate operating regime for solvent
trast to the bubble fractionation results where the sublation. In part 3 of this series, we will present
performance suffered as the diameter increased. It is steady-state fractional removal data collected from a
generally known that increasing the column diameter continuous, pilot sublation column and a bubble frac-
adversely affects the axial mixing in bubble fraction- tionation column.
ation (Kown and Wang, 1971; Deckwer et al., 1974;
Shah et al., 1982); however, it was concluded that, once Mass Transport Mechanisms
a solvent layer captures the solute as in sublation, it is
retained as the solvent possesses a greater affinity for In solvent sublation, there are three principle path-
the solute. From preliminary studies using the 15 cm ways or transport mechanisms available by which a
diameter column, it was recognized that the inability pollutant may be removed from the aqueous phase. They
to produce small bubbles (<1 mm) at high air flow rates are transport by air bubbles, water entrainment due to
(>600 mL/min) can be a problem in scaled-up operation. the wakes of rising air bubbles, and molecular mass
It was observed that, over a long duration, the activity transport across the solvent-water interface. These
in the stagnant solvent phase becomes greater than that mechanisms are illustrated in Figure 2a and discussed
in the water phase, in which case, the driving force for below.
mass transport is reversed. It was recommended that Transport by Air Bubbles. The most significant
future studies should investigate ways to produce and pathway for the removal of pollutants is the unidirec-
maintain small bubbles at high air flow rates and reduce tional transport by air bubbles. As air bubbles rise up
backmixing from the organic solvent to the aqueous the column, both volatile and nonvolatile compounds
phase by continuously replenishing the solvent phase. partition into and on the surface of the bubble. For
These results have motivated our group to further dilute systems of volatile compounds, Henry’s law
investigate aspects of solvent sublation at the pilot-scale. adequately describes the extent of partitioning and,
These aspects include (i) the development of a steady- ultimately, the extent of stripping. However, for the
state mathematical model more suited for engineering case of strongly hydrophobic compounds that are of low
applications and scaleup; (ii) the effect of novel bubble volatility such as polyaromatic hydrocarbons, chlori-
spargers on the bubble size and other hydrodynamic nated pesticides, etc., the partitioning process is pri-
properties, and (iii) the effect of air, solvent, and water marily due to interfacial adsorption upon the bubble
flow rates on the performance of a three-phase continu- surface. The extent of partitioning can be determined
ous sublation column. The compounds chosen to be by combining the equilibrium relationships for bulk-
investigated in this research are pyrene and penta- phase partitioning and surface adsorption with a mass
chlorophenol. These pollutants have unique physio- balance for the total amount of pollutant carried by a
chemical properties as shown in Table 2. bubble of radius a. In eq 1, m is the total amount of
In part 1 of this series, we discuss the mechanisms pollutant, Γ is the surface concentration of the bubble,
of transport in solvent sublation and develop math- and Cv is the concentration of pollutant in the air
+ +
m
CA )
4 3
πa
) (a3K
A )
+ Hc Cw (3)
3
The term in parentheses is the effective Henry’s law
constant, H. It depends inversely upon the air bubble
size. The ratio of the Henry’s law constants, H/Hc, is
termed the enhancement factor, H. From the values
of Hc and KA shown in Table 2 for pyrene, one can
determine that H is at least an order of magnitude
greater than Hc. When the bubble radius is on the order
of 0.05 cm, H becomes over a 100 times greater than
Hc.
Bubble-Wake Entrainment. Bubble-wake en-
trainment is the second most important transport
mechanism in solvent sublation. It results from the fact
that a bubble rising in a water column is not exactly
spherical. It is actually shaped more like a semispheri-
cal cap. Located behind the bubble resides a small
quantity of water called the wake (Fan and Tsuchiya,
1990), the amount of which is defined by the streamlines
in the fluid created by the rising bubble. When the
bubble and wake enter the solvent, the wake disengages
and becomes a small droplet that slowly descends to the
aqueous phase. This releases pollutant back into the
solvent. The droplet returns to the aqueous phase with
an equilibrium amount of pollutant. To account for this
type of transport, it is assumed that the wake uniformly
surrounds the bubble such that the volume of the wake
can be approximated as 4πa2di, where di is the wake
thickness.
Solvent-Water Interfacial Molecular Mass
Transfer. Unarguably, the organic solvent-water
interface is the focal point for all mass transport. As
shown in Figure 2a, one of the transport mechanisms
across the interface is molecular mass transport. In a
conventional liquid-liquid extraction process, this is the
sole mechanism of transport and is very important.
However, it is not as important in solvent sublation. The
explanation for this is that, in a sublation column,
convective mass transport due to the rising air bubbles
overwhelms the rate of molecular mass transport and
prevents the establishment of equilibrium across the
solvent-water interface. This is only true if the sub-
lation column is operated in a three-phase continuous
mode where steady-state operation is possible.
Model Development
model and the axial dispersion model to solvent subla- persion coefficient and a second derivative. In this
tion. We further show that one can easily convert from development, axial dispersion in the gas phase is also
one model to the other. included.
Series CSTR Model (SCM). A general illustration Figure 2c shows a depiction similar to that used in
of the SCM appears in Figure 2b. The top stage, the SCM; however, the transport mechanisms within
designated as stage 1, represents the thin layer of the column have been drawn about a differential ele-
solvent which floats upon the column of water. The ment, dx. The solvent is not considered part of the
stages below the first make up the water column, which continuum and is treated as a separate well-mixed box.
is partitioned into an air phase and a water phase. The The governing equations that describe the transport of
volume fraction of the air phase is equal to the gas pollutant are
holdup, , while that of the water phase is 1 - . The
total number of stages comprising the water column ∂A ∂2A ∂A
determines the amount of axial mixing present in the ) 2 + Pe1 + Sh1(W - A) (10)
∂τ ∂ξ ∂ξ
column. The simplest case to consider is a two-staged
model (or two-box model); i.e., the water column is
comprised of just one stage (box). ∂W ∂2W ∂W
) 2 - Pe2 - Sh2(W - A) (11)
The transport mechanisms shown in Figure 2b in- ∂τ ∂ξ ∂ξ
clude convection, molecular-mass transfer between the
water and solvent, molecular-mass transfer between the 1 dS
) PA(HA|ξ)0 - S) + (Eo + Sto)(W|ξ)0 - S) - S
water and air, and bubble-wake entrainment. These φ dτ
mechanisms are included in steady-state mass balances (12)
that describe the transport of pollutant in the sublation
Equations 10 and 11 apply to the air and water phases
process. For a two-box SCM, the governing mass
and are referred to here as the two-phase axial disper-
balances are
sion model (ADM2). Equation 12 is a mass balance
( )
around the solvent layer. When the time derivatives
3k 2 CA in eqs 10-12 are set to zero and the appropriate
QwCw° - QwCw - πrc L Cw -
a H
-
boundary conditions are used, a steady-state solution
( )( )
di Co can be found as shown in the appendix. In the general
πrc2k1(1 - ) + 3QA Cw - ) 0 (4) case, the Peclet and Sherwood numbers would be
a Kow written in terms of two different axial dispersion coef-
ficients, one for the water phase and one for the air
QACA° - QACA +
3k 2
a (
πrc L Cw -
CA
H
)0) (5)
phase. Since it will be shown in part 2 of this series
that the bubbly flow regime is the preferred regime for
solvent sublation, the gas-phase dispersion coefficient
( )
is assumed to be equal to that of the liquid phase. This
HcCo is a good assumption for the bubbly flow regime;
QoCo° - QoCo + QA CA -
Kow
+
however, it has been reported (Shetty et al., 1992) that
( )( )
di Co the mechanism of gas-phase dispersion in the hetero-
πrc2k1(1 - ) + 3QA Cw - ) 0 (6) geneous regime (characterized by two distinct bubble
a Kow sizes) is distinctly different from that in the homoge-
neous regime. Therefore, the single-dispersion-coef-
where Cw, CA, and Co are the effluent water, air, and ficient assumption is not valid in the heterogeneous
solvent molar concentrations, respectively, and Qw, QA, regime.
and Qo are the water, air, and solvent volumetric flow A relationship exists (Fogler, 1992) that allows one
rates, respectively. The superscript degree, designates to determine the axial dispersion coefficient used in the
the molar feed concentration; however, in this analysis, ADM2 from the number of well-mixed boxes used in the
it is assumed that the air and solvent enter the process SCM. The derivation of the relationship involves de-
free of pollutant (i.e., CA° ) Co° ) 0). When eqs 4-6 termining the effluent response curves (or residence
are divided through by QwCw°, QAHCw°, and QoKowCw°, time distributions, RTDs) of both models when the
respectively, the following dimensionless equations mass-transfer terms are set to zero, and the feeds to
result: each model are subjected to a perfect impulse. Next,
the variance (second moment) of the RTD for n-CSTRs
1 - W - StwA(W - A) - (Stw + Ew)(W - S) ) 0 (7) in series is set equal to the variance of the RTD for the
axial dispersion model. The result is eq 13. After
A° - A + StA(W - A) ) 0 (8) 1 2 8
(13)
n Pe1 Pe 2
) +
1
S° - S + (Sto + Eo)(W - S) + PA(HA - S) ) 0 (9)
rearrangement, the expression for the dispersion coef-
The system of eqs 7-9 has an analytical solution which ficient is obtained as
is given in the appendix. When the number of boxes
used in the SCM is greater than two, a numerical uL
D) (14)
algorithm is recommended. n + (n2 + 8n)1/2
Two-Phase Axial Dispersion Model (ADM2). An
alternative method for modeling the sublation process where n is the number of CSTRs making up the water
is to consider the water column as a continuum of well- column (n ) i - 1), u is the gas velocity (ug/), and L is
mixed, interacting boxes. The interactions allow for the length of the vessel. The obvious benefit of this
axial mixing which is defined in terms of a dis- relationship is that one can easily convert between the
+ +
Figure 3. Fractional removal versus QA and Qw as predicted by Figure 4. Fractional removal versus QA/Qw shows multiple
the SCM. steady-states.
Table 3. Model Parameters Geary and Rice, 1991). The bubble-wake correlation
thermodynamic kinetic hydrodynamic design appearing in the table possesses the same functional
Hc ) 5 × 10-4 k ) 5 × 10-4 cm/s a ) 0.1ug1/2 cm rc ) 5.08 cm form of experimental data cited elsewhere (Fan and
KA ) 5 × 10-3 cm k1 ) 10-4 cm/s ) 0.05ug L ) 152.4 cm Tsuchiya, 1990). Overall, the characteristics of the
Kow ) 76412 n)1 Vw/Vb ) 1.5 × bubbly flow surface plot are what one would expect: fractional
10-4Vb-0.741 type sparger removal varies directly with air flow rate and inversely
with water flow rate; and the approach toward unity is
two models given either the number of CSTRs or the asymptotic. Though the surface plot nicely illustrates
axial dispersion coefficient. the model, it is not accessible for quantitative use.
Furthermore, multiple surface plots are required to span
Results and Discussion Qo. Because of these reasons, an alternate way to
present the model is necessary.
The two response functions that describe the ef- Since it has been customary in previous sublation
fectiveness or performance of a sublation column are studies to plot fractional removal versus a ratio of flow
the fractional removal of pollutant, FR, and the separa- rates, the same fractional removal data used in the
tion factor, Σ, previous figure was plotted versus QA/Qw (see Figure
4). As can be seen in the figure, multiple steady states
Cw are predicted. The reason why this occurs is that
FR ) 1 - W ) 1 -
Cw° fractional removal is affected by column hydrodynamic
properties which are functions of gas velocitysnot QA/
S Co Qw. For example, the hydrodynamics of the column are
Σ) (15) different at (QA ) 2000 cm3/min; Qw ) 20 cm3/min) than
W CwKow
)
they are at (QA ) 10000 cm3/min; Qw ) 100 cm3/min)
even though the ratios of QA/Qw are the same. If Qw
The separation factor is the ratio of the effluent solvent were fixed, then multiple steady states would not occur
concentration to the effluent water concentration and as the ratio QA/Qw would be directly proportional to gas
is a good measure of the effectiveness of solvent subla- velocity. Again, the problem with fixing a variable is
tion relative to other liquid-liquid separation processes. that multiple plots become necessary. Therefore, it was
Both response functions are dependent upon five types decided to correlate the model predictions in terms of
of variables: thermodynamic, kinetic, design, opera- the dimensionless numbers that describe the mass
tional, and hydrodynamic. Not all of these variables, transport mechanisms. The advantage of this approach
however, are independent. In most cases, the kinetic is that the effect of the operational variables and the
and hydrodynamic variables are defined in terms of the effect of their influences on column hydrodynamics are
gas velocity, which is an operational variable. Hydro- combined. Based on multiple-variable regression analy-
dynamic properties are also dependent upon the design ses of 250 computer simulations, correlations for frac-
variables, especially the sparger type. Thus, the prob- tional removal and the separation factor were deter-
lem of characterizing sublation column performance is mined. The ranges of the operational variables used
one which investigates the effects of the operational, in the simulations were 400 e QA e 15 200 cm3/min,
design, and thermodynamic variables. 20 e Qw e 100 cm3/min, and 0.1 e Qo e 100 cm3/min.
Effect of Operational Variables (QA, Qw, Qo). In For strongly hydrophobic compounds where the solute
Figure 3, a representative surface of fractional removal capacity of the solvent is large (Pw e 0.013), a simple
versus QA and Qw is shown for a strongly hydrophobic correlation for the fractional removal was found.
compound, like a polynuclear aromatic hydrocarbon
(PAH). The surface was generated from the SCM using FR
typical values (i ) 2 and Qo ) 2 cm3/min). An equiva- ) Ew0.1657StwA0.8352 (16)
1 - FR
lent surface would result from the ADM2 if Pe1 ) 4.
Other model parameters that were used to generate the In Figure 5a, a parity plot of fractional removal shows
surface are listed in Table 3. As shown in the table, very good agreement between the theory and the
the hydrodynamic variables were determined from correlation, as all the deviations are within (10%. The
power-law models which are representative of the reason for the good agreement is that the functional
bubbly flow regime (Clift et al., 1978; Shah et al., 1982; dependence of eq 16 is consistent with the relative
+ +
ΨKow-1Qo-1
Σ∼ (19)
1 + ΨKow-1Qo-1
where
[ ]QAdi 1.464
Ψ≡ 3 [QAHc]0.544 (19)
a
The physical importance of eq 19 becomes clear when
one recognizes that the terms remaining in the brackets
(collectively referred to as the kinetic factor, Ψ) reflect
the kinetic and hydrodynamic processes associated with
the air (bubble) phase. Essentially, Ψ controls the
extent to which equilibrium is approached between the
solvent and water. First, in the limit of small Ψ, the
departure from equilibrium is great, and as eq 19
suggests, Σ is predicted to be zero. Of course, a zero
separation factor is not accurate since the physics of the
process change significantly when the air phase is
removed. The lower limit is better described by con-
sidering the solvent and water phases as two contacting
CSTRs. Though such a situation is kinetically undesir-
able in practice, it does represent a theoretical lower
limit for solvent sublation and does exist for a single-
staged quiescent solvent extraction process. It can be
Figure 5. Parity plots of the theoretical (a) fractional removal shown that, for such a case, Σ is equal to Sto/(1 + Sto).
and (b) separation factor versus predictions by correlation. Using this expression and typical conditions used in this
paper (Sto ) 6.34 × 10-6), one can estimate that the
importance of the mass-transfer mechanisms. The mass lower limit for Σ ≈ Sto , or equivlalently Σ/(1 - Σ) ≈ Sto.
transfer associated with the air bubbles, StwA, is weighted As one can see in Figure 5b, predictions of Σ for practical
more than that associated with entrainment, Ew. As sublation conditions are as much as 3 orders of magni-
alluded to, the mass transfer across the solvent-water tude greater than the lower theoretical limit. The
interface (Stw or Sto) has no effect on the fractional improved separation is attributed to the additional
removal for systems which are overwhelmed by the transport processes associated with the air (bubble)
transport of air bubbles. One advantage of the correla- phase. In the limit of large Ψ, the separation factor
tion is that fractional removal of strongly hydrophobic approaches unity, the condition for solvent-water equi-
compounds can be estimated very easily without resort- librium. In practice, however, it is unlikely that equi-
ing to the analytic solution. Another advantage is that librium may be reached solely by adjusting Ψ. Since
the correlation suggests the appropriate way in which the hydrodynamic properties were assumed to be that
experimental data should be plotted, as will be shown of homogeneous flow, the SCM will eventually break
in a later paper (part 3 of this series). down as Ψ becomes very large. Thus, other means of
From the regression analysis, a simple correlation for improving the performance are needed, such as reducing
the separation factor was also obtained. the solvent usage, Qo, without significantly affecting the
overall fractional removal.
Σ It is emphasized that the correlations, eqs 16 and 17,
) Eo1.464PA0.544QoKow (Qo in cm3/min) (17)
1-Σ were developed from the SCM with fixed kinetic pa-
rameters (see Table 3) and are valid for strongly
A parity plot of this correlation and the theoretical hydrophobic compounds (Pw e 0.013). For weakly
separation factor is shown in Figure 5b. Again, the hydrophobic compounds, the exact solutions given in the
agreement is within (10%. One interesting feature appendix must be used.
about the correlation is that, though it is appears first Effect of Thermodynamic Variables. Examining
order in QoKow, the overall dependence is actually the effects of thermodynamic properties is important
≈(QoKow)-1. To explicitly see the inverse dependence, because they can vary greatly between families of
eq 17 is rearranged and expressed in terms of dimen- hydrophobic pollutants. As shown in Table 2, the
sional numbers. interfacial partition constant can vary over 4 orders of
magnitude. Other properties such as the Henry’s law
Co
3 {
QoKowa } { }
QAdi 1.464 QAHc 0.544
QoKow
QoKow
constant and the solvent-water partition coefficient can
vary greatly as well. Therefore, computer simulations,
{ } { }
Σ) again based on the SCM (i ) 2) and values cited in Table
CwKow QAdi 1.464 QAHc 0.544
)
3, were performed to determine the relative importance
1+ 3 QoKow of these thermodynamic properties.
QoKowa QoKow
(18) There was negligible effect of the solvent-water
partition constant on fractional removal for Kow g 700.
Observing that the exponents for the terms in braces This was not a surprise given the fact that fractional
sum to approximately 2, one can factor a (QoKow)-2 from removal was shown not to be dependent upon solvent-
+ +
[ ][ ( )]
kπrc2(1 - )/QoKow Ew + Stw StA
Stw ) water-side Stanton number (solvent-water), Stw ≡ 1 + PA 1 - H
klπrc2(1 - )/Qw 1 + PA + Sto + Eo 1 + StA
+ +
Eo + Sto + PAH
StA
[( ) ]
A ) λ4 1 +
λ4
Pe1
- eλ4ξ -
[( ) ] [( ) ]/
Σ 1 + StA
( )
λ2 λ3
1-Σ StA κ 2 eλ2 1 + - eλ2ξ + κ3 eλ3 1 + - eλ3ξ
)
1 + PA 1 - H Pe1 Pe1
[ ]{[ ( ) ]
1 + StA
γ1 - γ2 λ4 β4(γ1 - λ4) - γ2
eλ4 1 +
Pe2 Pe1
- -
γ1 - γ2
[( ) ]
The appropriate boundary conditions for the ADM2
are λ2 β2(γ1 - λ2) - γ2
κ2 eλ2 1 +
Pe1
- +
γ1 - γ2
dA
|
dξ ξ)0
)0
dW
|
dξ ξ)1
)0 [( ) κ3 λ3 1 +
λ3
Pe1
-
β3(γ1 - λ3) - γ2
γ1 - γ2 ]}
dA
| ) -Pe1Aξ)1 [( ) ]
W ) eλ4 1 +
λ4
Pe1
- β4eλ4ξ -
[( ) ] [( ]/
dξ ξ)1
κ2 eλ2 1 +
λ2
- β2eλ2ξ + κ3 eλ3 1 +
λ3
)
- β3eλ3ξ
|
Pe1 Pe1
[ ]{[ ( ) ]
dW
) Pe2(Ew + Stw)(W|ξ)0 - S) - Pe2(1 - W|ξ)0) γ1 - γ2 λ4 β4(γ1 - λ4) - γ2
dξ ξ)0 eλ4 1 +
Pe2 Pe1
- -
γ1 - γ2
where it has been assumed that the solvent and air
enter the column free of pollutant. The analytic solution [( )
κ2 eλ2 1 +
λ2
Pe1
-
β2(γ1 - λ2) - γ2
γ1 - γ2
+ ]
is given in terms of the following variables:
Pe1 1 2
[( ) κ3 eλ3 1 +
λ3
Pe1
-
β3(γ1 - λ3) - γ2
γ1 - γ2 ]}
β2 ) 1 - λ - λ where
Sh1 2 Sh1 2
x ( )( ) ( )( )
a1 λ4 β4eλ4 - β3eλ3 λ4 β4eλ4 - β2eλ2
λ2 ) 2 -Q cos (3θ + 240°) - 3 κ2 )
λ2 β eλ2 - β eλ3
κ3 )
λ3 β eλ2 - β eλ3
2 3 2 3
Pe1 1 2
β3 ) 1 - λ - λ and
Sh1 3 Sh1 3
x
a1 HPAA|ξ)0 + (Eo + Sto)W|ξ)0
λ3 ) 2 -Q cos (3θ + 120°) - 3 S)
1 + PA + Eo + Sto
x
Pe1 a1 Literature Cited
1 2
β4 ) 1 - λ - λ
Sh1 4 Sh1 4
λ4 ) 2 -Q cos (3θ) - 3 Akita, K.; Yoshida, F. Gas Holdup and Volumetric Mass Transfer
Coefficient in bubble Columns. Ind. Eng. Chem. Process Des.
Dev. 1973, 12, 76.
a1 ) Pe1 - Pe2 Caballero, M.; Cela, R.; Perez-Bustamante, J. A. Solvent Sublation
of Some Priority Pollutants (Phenols). Anal. Lett. 1988, 21 (1),
63-76.
a2 ) -(Pe1Pe2 + Sh1 + Sh2) Caballero, M.; Cela, R.; Perez-Bustamente, J. A. Studies on the
Use of Colloidal Gas Aphrons in Coflotation and Solvent
Sublation Processes. A Comparison with the Conventional
a3 ) Sh1Pe2 - Sh2Pe1 Technique. Sep. Sci. Technol. 1989, 24, 629-640.
Cervera, J.; Cela, R.; Perez-Bustamante, J. A. Analytical Solvent
Sublation of Metallic Dithizonates Part I. Solvent Sublation of
cos θ ) R/x-Q3 Copper. Analyst 1982, 107, 1425-1430.
Clift, R.; Grace, J. R.; Weber, M. E. Formation and Breakup of
Fluid Particles. Bubbles, Drops, and Particles; Academic
R ) (9a1a2 - 27a3 - 2a13)/54 Press: San Diego, 1978; pp 321-328.
Deckwer, W. D.; Schumpe, A. Improved Tools for Bubble Column
Reactor Design and Scale-Up. Chem. Eng. Sci. 1993, 48 (5),
889-911.
Q ) (3a2 - a12)/9 Deckwer, W. D.; Burckhart, R.; Zoll, G. Mixing and Mass Transfer
in Tall Bubble Columns. Chem. Eng. Sci. 1974, 29, 2177-2188.
[ ( )]
Deckwer, W. D.; Tien, K. N.; Kelkar, B. G.; Shah, Y. T. Applicabil-
Eo + Sto ity of Axial Dispersion Model to Analyze Mass Transfer Meas-
γ1 ) Pe2 1 + (Ew + Stw) 1 - urements in Bubble Columns. AIChE J. 1983, 29 (6), 915-921.
1 + PA + Eo + Sto
Fan, L. S.; Tsuchiya, K. Wake Sizes. In Bubble Wake Dynamics in
Liquids and Liquid-Solid Suspensions; Brenner, H., Ed.;
PAH Butterworth-Heinemann: Stoneham, MA, 1990; Chapter 5.
γ2 ) Pe2(Ew + Stw) Finch, J. A.; Dobby, G. S. Column Flotation; Pergamon Press:
1 + PA + Eo + Sto Toronto, 1990; Chapter 1.
+ +