Chemical En@neerimg Science, V-d 45, No. 9, pp. MOZ-3CQ3,1990. ace-2so9/90 s3.00 + 0.

00
Printed in Great Britain 01990l’er~onPresspk

Optimal pore size of catalysts for hydrodemetallation reactions

(First received 21 August 1989; accepted in revised form 22 March 1990)

INTRODUCTION section method, the pore sizes which yield the optimal initial
Pore size optimization is essential to the design of hydro- activity and the optimal-lifetime activity can be calculated.
demetallation catalysts. Rajagopalan and Luss (1979) de- Figure 1 plots l/n: YS 4 with reaction order as the para-
veloped some simple formulas to determine the optimal pore meter. The results from Fig. 1 indicate that the reaction
size for initial activity and the optimal pore size for total order affects the optimal l/Q only slightly. For values of 4
lifetime activity. Their studies considered that the rate of between 1 and IO, which are typical values in industrial
demetallation was first-order with respect to the metal con- practice (Limbach and Wei, 1988), the differences in l/n:
centration. However, besides the first-order reaction, some among three reaction‘ orders are no more than 8%.
other authors (Beuther and Schmid, 1963; Oleck and Sherry, From the viewpoint of industrial applications, the selec-
1977) observed second-order deposition kinetics. In addi- tion of optimal pore size is better determined from the total
tion, Hung and Wei (1980) used the model compound to lifetime activity. Here the total lifetime activity is calculated
study the hydrodemetallation kinetics. Their results indi- by the summation of catalyst activity from the beginning to
cated that the reaction order concentrated in the neighbor- an age which yields the limiting activity. The optimal l/A:
hood of one-half. In fact, from the experimental result, the for different reaction orders are presented in Fig. 2. The
demetallation reaction is first-order for most of the reaction parameter, F, in Fig. 2 represents the ratio of limiting to
range from the initial period, and, due to the complicated optimal initial activity as defined by Rajagopalan and Luss
constituents of residium oil, the unreacted portions may (1979). In contrast to the results of Fig. 1, the differences in
deviate from first-order in the last part of the reaction. l/Q among three react&n orders are significant, and may
Therefore, it is of value to examine the effect of reaction order reach 25% for values of 4 between 1 and 10. This means that
on the optimal pore size in hydrodemetallation reactions. the intrinsic kinetics of hydrodemetallation reaction should
be identified before determining the optimal pore size of the
MATHEMATICAL MODEL hydrodemetallation catalysts. Although only slab pellets are
discussed here, similar behavior for spherical pellets would
The mass balance of metal-containing molecules in a
bc observed.
single cylindrical pore for an nth-order reaction can be
expressed as
C.-L. CHIANGt
C. H. YU
(1) Department of Chemical Engineering
National Cheng-Kung University
df u”u, Tainan, Taiwan 70101, R.O.C.
-- (2)
t= r,(O) NOTATlON
where C concentration inside the pore, kmoI/m”
I.
(3)
‘Author to whom correspondence should be addressed.

(4)

Boundary conditions and symbols in eqs (l)-(4) are the

same as those proposed by Rajagopalan and Luss (1979). In - Ist-order
order to compare the optimal pore sizes under different
reaction order kinetics, the following assumption is made, i.e.
k, = k2Co = k,,,C,““. (3
Then the same Thiele modulus and metallation velocity, v,
are obtained, i.e.

41 = 62 = 41/z (6)
01 = uz = VI,*. (7)
In this analysis, a slab catalyst with uniform pores is
adopted to investigate its activity performance. The concen-
tration profile of metal-containing molecules in the pore can
be solved numerically from eqs (1) and (2). Then the de-
metallation rate per unit volume of the slab, R,,, can be
calculated. In this study, R,, has the same form as eq. (19) of
Rajagopalan’s paper.
10-l 1 10 10Z
RESULTS AND DISCUSSION
7
In this analysis, the same data as proposed by
Rajagopalan and Luss (1979) are used. By using the golden- Fig. I. Dependence of l/n,+ on 6 for a slab catalyst.

3002
 Shorter Communications 3003

r&,(O) initial pore radius, m

R,, demetallation rate per unit volume of slab,
kmol/m3-s
- 1 st -order
--- - Pnd-order U dimensionless concentration (C/C,)
60 - - - half -order V metallation velocity, m/s
z distance along pore, m

Greek letters
1 ratio of molecular to pore radius
3, for which optimal initial activity is attained
:; A for which optimal lifetime activity is attained
dimensionless distance in pore (z/l)
f Thiele modulus defined by eq. (4)
Thiele modulus defined by eq. (3)

Subscripts and superscripts

n nth-order reaction
1 first-order reaction
2 second-order reaction
-lo-’ 1 10 10* l/2 half-order reaction

REFERENCES
Fig. 2. Dependence of l/A: on & for a slab catalyst Beuther, J. and Schmid, 3. K., 1963, Proceeding of the
(F = 0.25). 6th World Petroleum Congress, Frankfurt: Verein zur
Forderung des 6. Welt-Erdol-Kongress, Hamburg; Sec-
tion 3, Paper 20, p. 297.
Hung, C. W. and Wei, J., 1980, The kinetics of porphyrin
CC. bulk concentration, kmol/m” hydrodemetallation. Ind. Engng Chem. Process Des. Deo.
D Al? bulk liquid diffusivity, m2/s 19, 250-263.
effective diffusivity in the pore Limbach, K. W. and Wei, J., 1988, Effect of nonuniform
$ dimensionless pore size [r,(z, t)/r,(O)] activity on hydrodemetallation catalyst. A.Z.CL.E. J. 34,
ratio of limiting to optimal initial activity 305-313.
G(A) (1 - A)4 Oleck, S. M. and Sherry, H. S., 1977, Fresh water manganese
g dimensionless diffusivity [D,(z, t)/D,(O)] nodules as a catalyst for demetallizing and desulfurizing
demetallation rate constant, m/s petroleum residua. Xnd. Engng Chem. Process Des. Dev. 16,
: half-length of pore 525-528.
n reaction order Rajagopalan, K. and Luss, D., 1979, Influence of catalyst
r, radius of molecule, m pore size on demetallation rate. Ind. Engng Chem. Process
rc pore radius, m Des. Dev. 18, 459-465.

Chemical Ellg*locr*yl Scirnce. Vd. 45, No. 9, pp. 300~3006, 1990. cloos2509/90 Em0 + 0.00
Printed in Great Britain. 0 1990 Pergalnon Press plc

Simulation of particle size distribution in an aggregation-breakup process

(First received 30 January 1990, accepted in revised firm for publication LO April 1990)

INTRODUCTION We propose a simple experimental procedure similar to

In the processes of particle flocculation using polyelectro-
lytes, the destabilization of the dispersion is attributed prim-
arily to the mechanisms of charge neutralization or polymer
bridging, producing floes of quite distinct features. Those
formed by polymer bridging are generally considerably
stronger than floes formed by charge neutralization
(Gregory, 1987). However, bridged floes broken by shear
may not be readily re-formed (Pelton 1981), probably be-
cause of scission or rearrangement of absorbed polymer
chains (Gregory, 1987; Stratton, 1983). When charge neutral-
ization becomes the dominant mechanism, floes broken by
shear can be reformed when shear is reduced. This property
has been used by Ditter et al. (1982) to distinguish the
mechanisms of flocculation.
that used by Ditter et al. (1982) to identify the mechanism of
flocculation. Destabilized particles are first allowed to ag-
gregate under gentle agitation until an almost steady particle
size distribution (PSD) is achieved as a result of the equilib-
rium between aggregation and breakup. Next the particles
are subjected to strong agitation for a time, and subsequently
are allowed to reaggregate under the original gentle agita-
tion. The kinetics of flocculation can he followed by
measuring the PSD at different stages. The mechanism of
charge neutralization is indicated by a rapid recovery of the
particle size after termination of the strong agitation, while
the bridging mechanism is indicated by a failure to exhibit
this recovery.
The use of the above criterion, however, is subject to one