Chaudhari1980 PDF
Chaudhari1980 PDF
Chaudhari1980 PDF
SCOPE
Three-phase reactors have many diverse applications and polymerization reactions. A recent application of
in catalytic reaction engineering. There are two common slurry reactors is in the field of pollution control, where
modes of operation of the three phase reactor: (1) trickle a gas or liquid phase contaminant is oxidized in a slurry
bed or packed bed operation where the catalyst is sta- catalytic reactor. The potential application of slurry reac-
tionary and the liquid flows as a dispersed phase, the tors is likely to expand with new developments in the
gas being the continuous phase and (2) slurry reactors field of %eterogenized' homogeneous catalysts. These cata-
where the catalyst is suspended in the liquid medium by lysts are in the form of polymer bound transition metal
either mechanical or gas-induced agitation. The slurry complexes, and the reactions are normally carried out in
is now the continuous phase and the gas is well dispersed a slurry reactor. (For example, hydroformylation of ole-
in the reactor. Here the liquid medium could either be fins, carbonylation of methanol to acetic acid, and oxida-
a reactant or an inert medium for contacting the dissolved tion of ethylene to vinyl acetate). Slurry operations have
gases with the solids. Similarly the gaseous component many advantages over other modes of three-phase con-
could be either a reactant or an inert to pravide agita- tacting, e.g., in controlling the temperature and reducing
tion. The solid particles in most cases are catalysts or the extent of intraparticle diffusion. Since these reactors
adsorbents. A number of reviews are available on trickle operate at moderate conditions, higher selectivity can
bed reactors (Satterfield 1975, Goto et al. 1977) and packed also be envisaged.
bed three-phase reactors (Hofmann 1978), while there is I t is the purpose of this review to evaluate the major
no recent review on the slurry reactors. engineering developments and design procedures in this
Slurry reactors are commonly used in many industrial area, and to discuss methods of analyzing slurry reactor
systems. The scope of the review is restricted to the cases
processes. Some notable examples are hydrogenation of where the solid is either a catalyst or an adsorbent. The
unsaturated oils, Fischer-Tropsch reaction for hydrocarbon situation where the solid undergoes a reaction in the
synthesis, oxjdation of olefins, ethynylation of aldehydes, slurry is not considered.
The sIuny reactors have a number of advantages over GENERAL THEORETICAL ANALYSIS
other three-phase reactors, such as tickle bed or packed A general three-phase slurry system can be represented
bubble bed reactors. These are: by the reaction scheme
1. As small particle size of the catalyst can be used
in a slurry reactor, the intraparticle diffusional resistance +
A VB Products (i)
is less in comparison to a trickle or packed bubble bed
reactor. The trickle bed reactors normally employ catalyst The species A is generally a reactant in the gas phase
particle sizes at which the intraparticle diffusion may be and B is a nonvolatile reactant in the liquid phase. The
significant. reaction of A and B is assumed to occur at the interior
2. The external mass transfer coefficients in shrry surface of the catalyst particles, which are suspended in
reactors are higher than in trickle or packed beds, which the liquid medium. A number of industrially important
leads to better utilization of the catalyst. reaction systems conform to this scheme. Examples are
3. SIurries have higher heat capacities and higher heat found in hydrogenation and oxidation reactions. In some
transfer coefficients. Due to this, temperature control of industrially important cases, both A and B may be present
exothermic reactions is better in slurry reactors, and the in the gas phase, Oxidation of SO2 over activated carbon
formation of hot spots can be avoided. Slurry reactors catalyst is an example which has applications in pollution
are relatively safer for reactions with temperature run- control. Methanation reaction and Fischer-Tropsch s p -
away. The large liquid volume is also an advantage in thesis are other well-known examples.
maintaining isothermal conditions. The heat recovery, The intrinsic rate of reaction over the active sites
too, in these reactors is better. of the catalyst per unit weight of catalyst may be repre-
4. In trickle bed reactors, partial wetting of catalyst sented by a power-law kinetic model
surface may exist for a certain range of liquid flow rates,
and in such cases the entire catalyst may not be utilized
n = kcm+n,AmBn (1)
in certain situations (see, for example, Ramachandran where, SL is the local rate of reaction at a point within
and Smith 1979). In slurry reactors, this problem is not the catalyst where the concentrations are A and B. The
encountered. analysis of the reaction rate in a slurry system is simplified
5. In view of the pelletizing difficulties and the high when one of the reactants ( B ) is in excess, The variation
cost involved in pelletizing, slurry reactors may prove to of B in the reactor is then not significant and concentra-
be more useful in some cases. tion of B is uniform throughout the catalyst, hence the
In spite of these several advantages of slurry reactors, kinetic model can be simplified as
they pose some practical problems. A major problem is
the ~djfficultyin separating the catalyst and handling of n = k,Am (2)
the slurry. Because of this, the application of slurry where, km is a pseudo mth order rate constant, equal to
reactors in continuous processes has been limited. k(m+,>Bln. A criterion for Equation ( 2 ) to be valid is
Two types of slurry reactor operations are normally discussed later.
encountered: mechanically agitated slurry reactors and An alternate and mechanistically more realistic way of
bubble column slurry reactors. The mechanically agitated representing the rate is in terms of the Langmuir-Hinshel-
reactors have the advantage of high heat and mass trans- wood (L-H) model, although the power law model has
fer efficiency, while they pose probIems like catalyst the advantage of mathematical simplicity. The reaction
attrition and complete backmixing of the liquid phase. In rate based on this model for single-site mechanism may
the bubble column reactors, catalyst is suspended by means be expressed as
of gas induced agitation. They have the advantage of low
power consumption. The backmixing can be slightly re- (3)
duced in this type of reactors. But they have a disadvan-
tage of nonuniform catalyst distribution and lower mass First the analysis of slurry reactors is made for the
and heat transfer efficiency. case when pseudo rnth order conditions are satisfied. The
This review considers two aspects of slurry reactors. effect of transport steps is analyzed in detail for reactions
The first is a mathematical development of the effect of of various orders and for L-H modeIs. A more general
mass transfer on several different intrinsic rate models case of (m, n)th order reaction is then analyzed.
in a slurry reactor. The second part is a literature review The analysis in this section is for differential conditions,
of correlations for predicting design parameters. The in- so that the conditions in the liquid phase are assumed to
formation presented here will be useful in selecting suit- change only slightly. This restriction is important because
able reactor type, and experimental evaluation of kinetic the pseudo-mth order rate constant, km, changes with
parameters and design. the concentration of liquid phase reactant B. Also, in
where
J
a The concentration of gas A leaving the reactor is
-
P
I ! I I i
R O
I Ag,=Agiexp(-cd) + H A A I [-~ e x p ( - a ~ L ) ] (8)
The rate of gas absorbed per unit volume of slurry is
- r
Figure 1. Concentration profile for a slurry catalyzed reaction.
TABLE REACTORINCORPORATING
RATE OF REACTIONIN A SLURRY
1. OVERALL EXTERNAL
MASSTRANSFERBUT IGNORING GRADENTS
CONCENTRATION
INTRAPARTICLE
1 1 -1
First-order T.1.1
A"[Ma+;;j;;l
Second-order T.1.2
MA MA
(wk1/212
Half-order T.1.3
2MA
Zero-order See text, Equations (35), (36) and (37)
Langmuir-HinseIwood
[single site]
- [(l+&A"+- wkl )2 - 4wk;;A' a'] } T.1.4
MA
Second-order
[two components] See Equation (48) in text -
A e E T.1.5
MA
2AeE
- [ ( A" + -
2wkz )
MA
- A'2 +A2 1" } T.1.6
n
as an independent step have also been proposed (Zwicky ?I 02
and Gut 1978). W
?
+
0
Rote o f Reaction Incorporating Intraporticle Diffusion
u.
Y 01
When the concentration of A is not uniform in the
catalyst particle, the above equations are not applicable. 2
a
A
RA = qcwkmA,"' (20) 03
[coth(34) - -1
1 1
?"=7 34 Figure 2. Overall effectiveness factor for a first-order reaction in a
slurry reactor.
where, 4 is the generalized Thiele modulus defined (Bis-
chaff 1965) as
and
as = AJA' (28)
Incorporating a, from Equation ( 2 6 ) and rearranging
For mth order reaction, 4, is given by ( 2 0 ) , the overall effectiveness factor, 7 for a mth order
reaction in a slurry reactor is given by
I 0
"
Overall Effectiveness Factor yw
Y
An effectiveness factor for a first-order slurry reaction ;o t
incorporating external and internal mass transfer effects I
0
1
has been defined by Sylvester et al. (1975). The concept
0 05
of an overall effectiveness factor is useful in simplifying
the calculation of rate of chemical reaction in a slurry
003
reactor. Ramachandran and Chaudhari ( 1979a) defined OI 0 2 03 05 10 20
the overall effectiveness factor as actual rate of reaction
divided by the rate based on the inlet gas Concentration
and neglecting all the transport resistances Figure 3. Overall effectiveness factor for a half-order reaction in a
RA slurry reactor.
7)=
wa(AQ) 10
v= wk, ( A* ) m
To obtain an analytical solution for 17 in terms of known
parameters, the surface concentration of A (A,) has to
be eliminated. This can be done by expressing the surface
concentration of A in terms of the overall effectiveness
factor itself. From Equation (11) we have u
7)
as=l-- (26) 0 04
where
=A 01 0 2 0 3 03
+o
I 0
=+ I*[ 20
i
30 5 0 10 20
MAA'
UA = Figure 4. Overall effectiveness fnctor for L-H type kinetics (single
wa(A*) site mechanism).
( 1 - 5)m-1]1'=
for concentration of A to be finite inside the catalyst is
$=-[
R ( m + 1)
3 2
ppk,A"m-'
DeA
given by
Agi > (4)
critical (33)
where
(30)
Equation (29) is an implicit expression for 7 and has
to be solved by a trial and error procedure, as 4 is also
a function of 7. Having obtained 7 from Equations (29) (34)
and (30), the overall rate for mth order kinetics can be
When condition (33) is satisfied, the rate of reaction is
predicted using Equation (24). Thus, the overall effec- given by
tiveness factor aids in obtaining the rate of reaction in
a slurry reactor that incorporates all transport resistances. RA = wko (35)
For a first order reaction, the Thiele modulus given #
IJ
If the conditions are such that Equation (33) is not
by Equation (30) reduces to satisfied, and the concentration of A drops to zero at
+=,[-I
R PPkl
DeA
?la
some point 1, as shown in Figure 5, then the rate of
reaction can be calculated by a trial and error procedure
from the following set of equations
which is independent of a,. Hence the rate of reaction RA = wko[l - ( A / R ) 3 ] (36)
can be expressed as and
Aoi f 1 1
Z
0
kdm Ilc T.2.1
Y
ko T.2.2
T.2.5
T.2.6
111
As a simplification we assume
n = kz (A2 - A,') (43)
where k2 = k3B1 and Ae2 = E I 2 / B I K R . The rate of
r e a c t h per unit volume of slurry is then
RA AdA + 2wk,
= MA [ H MA - { ($ + z
y
- (-$)2 + A e 2 } "] (44)
-p
O = kl(A -Ae) (39)
where A, is the equilibrium concentration defined as
E l F i / B [ K R and kl the pseudo first-order rate constant
= k2B1. (It has been assumed here that the order of B" I
reaction with respect to A, B, E and I; is unily). The
rate of reaction per unit volume of slurry in the absence of I
significant intraparticle gradients (that is, wlieii the con-
dition given by Equation 17 is satisfied with m = 1) is I I
I a!
RA = klw ( A , - A,) (40)
Eliminating A, using Equation (11) we get an expression
for overall rate of reaction
R 0
In hydrogenation of fatty acid esters to fatty alcohols
(Muttzall and van den Berg 19682, the reaction is second- -r
order with respect to hydrogen and alcohol, and first-order Figure 7. Concentration profile for a general (m,n)th order slurry
with respect to ester. The reaction scheme for this case is reaction.
- 4v
A*BoM A M B
} "1 (48)
(58)
Comparing Equations (57) and ( 8 ) , the only difference
between the plug flow model and the complete mixing where the reaction orders are (ml, nl) and (mz, nz)
model is that the term [exp ( - ( Y A L ) ] is replaced by + +
for A B and A C reactions respectively.
+
1/(1 aAL). All the equations described earlier thus apply The selectivity defined as the ratio of the net rate of
formation ot C to the net rate of formation of E is then
also when the gas phase is completely backmixed, pro-
vided the above substitution is made. Comparing the given by
terms containing (YALin parameter M A for the two ex-
treme cases of plug flow and complete mixing. [1 -
exp( - - c u A L ) ] and 1 - 1/(1 +
( Y A L ) , respectively, both
the terms tend to ( Y A L when (YAL+ 0 (sparingly soluble
gas), and 1 when ( Y A L -+ 00 (highly soluble gay). Thus
the influence of gas phase mixing is negligible for these
extreme cases of gas soluliilities and may be important The influence of the external mass transfer parameter M A
only when ( Y A L is in the range of 0.5 to 10 and when on selectivity can be computed using Equation (59),
gas-liquid mass transfer is important. Similar considera- with RA calculated from Equation (58). When the order
tions would apply to other gaseous species, when two of reactions are different, a drop in the concentration
gases are reacting in the slurry reactor, of A due to mass transfer limitations will favor the reac-
tion of lower order, Thus if ml > m2 lower mass transfer
CONSECUTIVE A N D PARALLEL REACTIONS rates will decrease the selectivity while if ml < m2, re-
duced mass transfer rates will improve the selectivity.
Complex reactions with consecutive or parallel steps If m, = m2, the selectivity is independent of external
are commonly encountered in slurry catalyzed reactors. mass transfer parameter M A . These effects have also been
The types of reactions can be generally classified as experimentally verified by Kawakami and Kusunoki ( 1975)
for the hydrogenation of chlorobenzene according to the
Type I: A + B 4 C (iv) scheme:
H2 f C&CI+ CtjHs HC1 + (ix)
Type 11: 3Hz f CaHs --j CsHiz (4
The reaction orders with respect to hydrogen in a slurry
reactor with 0.5% Pt-carbon catalyst are 0.2 to 0.5
+A +A for the first reaction and 0.6 for the second reaction.
Type 111: B+C+E (viii) As in this case ml < m2, selectivity would decrease as
1 +2A T the stirring speed is increased, also as confirmed by
experiments.
!
0
5 a t 4 M C H 2 0 concentration -
0,
w
n
E
::4 t 4 m
m
n -0
0,
E M C H g concentration
0
\
E
-0 0
3
Di
.
)
E 0 3
r-
.
0 9
X
:
X
a
4
4 2 2
a
t I
0
5 10 15 20 O, 0.2 0.4 0.6 0.8 4 *o
PRESSURE, atm PRESSURE, a t m
Figure 10. A plot of RA vs partial pressure of hydrogen Car hydro- Figure 11. A plot of RA vs partial pressure of acetylene for ethynyla-
genation of DNT (zero-order reaction).
=
-
= th-oretical; 0 tion of formaldehyde over Cu-acetylide catalyst (zero-order reaction).
experimental data of Acres and Cooper, (1972). 0, =
0 experimental data of Kale and Chaudhari (1978).
Iu
(gas) at 1 atm. (101.32 kPa) ; kLa = 0.107 sec-l; w =
5.71 x g/cm3; ksap = 0.186 sec-'. .
P
B
HA
= + B;Q +(Ti p
clrs) g ~ (75)
1976b) follow zero-order kinetics with respect to B. In
this case the rate is given by where p g A is the partial pressure of A in the inlet gas
stream, 71, r2, 73 are empirical constants, and Bl and 'C1
= klA (70) are, respectively, the concentrations of reactant and prod-
uct, g mole/cm3.
The rate of reaction of B will then be given by Concentration of the product C1 can be related to the
concentration of reactant Bt by a stoichiometric mass
-
dBz vAgi 1
-
dt
= vRA =-
HA [
$!!L [1- exP(--crAl)]
balance
cc = C b +-VP
V
(Bl, - B1) (76)
-1
+- 1 + A(71)
] lowing
Substituting Equation (76) in (74), we obtain the fol-
correlation for solubility in terms of B1
ksa, llcwk1
The effectiveness factor T c in Equation (72) is given by When reaction is first-order with respect to A and B,
Equation (21), with the term 4 defined by Equation the rate of change of BE with time (incorporating the
(31). I n some cases, it has been observed that the rate is changes in solubility values of A as a function of Bl and
independent of concentration of B at higher concentra- ignoring intraparticle diffusion effects) is
tions but shows first-order dependence when B is less
than a critical concentration Btc, To model the reactor
performance for such systems, Equation (72) has to be
used for Blo to Blc, and then Equation (69) must be
used for Bt, to Blf.
Cm,nlth Order Reaction
The rate of reaction of B per unit volume of slurry Integrating Equation (78), we get an expression for B
is now given by as a function of time, incorporating the effects of solu-
bility changes
Goto and Smith (1978) modeled a slurry reactor in Figure 13. Minimum agitation required for complete suspension for
which two gases react catalytically, assuming the reaction various catolyst loadings.
In 0
-
U
I I I
0 2 6 8 10
SUPERFICIAL GAS VELOCITY, u g crnhac
a=-- 6% (99)
&
5 10 20 30 A knowledge of gas hold-up and bubble diameter is
SPEED OF A G I T A T I O N , N r p r necessary to estimate ( / < L a ) in bubble columns. These
were correlated by Akita and Yoshida as
Figure 15. Comparison of correlations for k~ a in an agitated reactor.
---- = Yogi and Yoshida (1975); -= Calderbank and
Moo-Young (1961).
dB
-=26[T]
dT
gdT2pL -0.5 gdT3pL2 -0.12
[-I PL2
-0.12
[A]
(100)
>3.5 x 10-2
(95)
%
' - o.20 [ gdT2PL 1"' [ gdT3pL2 ]'"" [
]
( 1 - %I4 ST PL2 dFdF
Nagata (1975) indicates that the power consumption
(101)
in slurries should be taken as P p , / p L 1 where p s is the
density of the slurry. This correction factor may not be The correlation of Kawagoe et al. is illustrated in Figure
significant for low catalyst loadings and for catalysts with 16, where kI.n is plotted as a function of the gas velocity
low densities. ug. The conditions used for this plot were: r l T = 10 cm,
Yagi and Yoshida (1975) proposed a correlation for P L = 1 g/cm3, pL = 8 x 10-3 gm/cm/sec, S T = 72
kLa in an agitated contactor dyne/cm, D = 2 x l o p 5cm/sec.
[ ~ ] d 1 2 = 0 . 0 ( 3 0[T]"5
d1"PL "" ] " [gl 'o] , [ dIN2 Experimental M e t h o d s
Liquid-side mass transfer coefficient, k L a can be deter-
PLD
mined by measuring the physical absorption rates or bulk
concentration of the dissolved gas as a function of time
in a batch reactor. For determining the true overall
liquid-side mass transfer coefficient, a system with neg-
The above equation does not require a knowledge of the ligible gas-side mass transfer resistance shoiild be wed.
power consumption rate and represents the data on COZ This can be done by selecting a sparingly soliil)le gas
absorption in glycerol-water at 30°C satisfactorily. (such as C O P ) for aqueous systems. Alternately a pure
The above two correlations are compared in Figure 15 gas can be used.
as a plot of kLa vs AT. The parameter values used were The concentration time data can lie analvzed I y well-
d T = 10 cm, d, = 5 cm, L = 10 cm, p L = 1 g/cm3,
known methods, to obtain 1cr.a. Techniques based on using
PL = 8 x 10-3 g/cm/sec, ST = 72 dyne/cm, ug = 0.2 a gas which undergoes a slow chemical reaction in the
cm/sec, D = 2 x l W 5 cmP/sec, p g = 1.2 x g/cm3. liquid can also he used, The details of this method are
In a bubble column reactor, the mass transfer coef- discussed by Danckwerts ( 1970). A1)sorption with reac-
ficient k L can be estimated using a relationship suggested tion of COP in sodium carbonate-liicarbonate Iniffer solu-
by Akita and Yoshida (1974)
--
k ~ -d o.5
D
~ yL [
DPL
]"'[ ~ ~
FL2
B ['"I
~ P &n2pL
L ~ ]3'8
ST
tion is controlled by mass transfer iinder certain condi-
tions, and this system could be used to evaluate kLa.
Mehta and Sharma (1971) used this technique to deter-
mine kl,a in agitated contactors.
(97) This value could also he o1,tained from the observed
Kawagoe et al. (1975) developed the following correla- rate of reaction in a slurry system, provided the rate of
tion for k L in a bubble column reaction is controlled by gas-liquid mass transfer. Aiany
hvdrogenation reactions on supported Pt catalysts are
fairly rapid, so that the rate is controlled by gas-liqiiid
mass transfer. For a first-order slurry reaction, a plot of
D PL
The correlation is valid when the density difference be- SUPERFICIAL GAS V E L O C I T Y , U p , Cm/..C
tween the liquid and solid is not large. Figure 18. Comparison o f correlations for k, in a bubble column.
Sano et al. (1974) measured solid-liquid mass transfer
coefficients in agitated vessels and bubble columns. They
- = Sano et a/. (1974); ----
( 1965).
= Kobayashi and Soito
Tempera- Pressure
Diffusing species Solvent/medium Catalyst ture “C (am.1
solute to pore diameter varied from 0.088 to 0.506. The (1977) is based on the principle of the well known
experimeutal data was correlated by Wicke-Kallenbach method. Other dynamic methods for
estimation of D , are reviewed by Hamachandran and
Smith ( 1978b) (see also Furusawa and Suzuki 1975).
20
Equation (61). The inherent assumption for the basis of 5
linearity of these plots is that k ~ ais independent of w .
The intercept of this line would correspond to VL/HAQ I0
[l - exp(--a~L)]-l or to l / k L a for sparingly soluble t
gases.
The slope of this line gives [apdp/6ks + l / n c k l ] . To
-?
.& 5
evaluate k, and k, separately, three approaches are pos- u
sible: w
w 3
1. The solid-liquid. mass transfer coefficient can be 0
estimated from suitable correlations. The value of ]c, can I2
n
2
be found from the slope by subtracting the contribution
of k,.
2. Experimental value of the slope can be got at I
different particle sizes. The term ppdp/6k, would change 0 001 0003 0005 001 002 003 005
provided intraparticle diffusion effects are absent. The Figure 19. Effect of d, on the slope of A*/RA vs l/w plot, for a
term l/& would change as dpl.Owhen the intraparticle first-order reaction.