2
:
s m
m
k
S
f
S
2
3
:
3
:
f
S
A
m
mol
C
S
at z = 0
At steadystate:
) (
A A
S
A
r N r
S
= =
`
(2)
where
) (
S
AS A m A
C C k N
C
=
`
(3)
s m
m
k
S
f
m
C
2
3
:
C
A
N
A
C
S
As
(C
A
at solid surface)
C
As
(C
A
in solid)
external
film
z
active centres
FLUID
l
0
Page 7 of 90
or:
= =
=
=
y P C
S
P
S
y
m m m
S
A A
A
m
S
A A
A
m
k P k C k
P P
N
k
y y
N
k
) (
) (
A
m S
m
S
A
S
A A m
S
A S
C
k k
k
C
C C k C k
C
C
S
S C S
+
=
= ) (
Substitute this back into (1):
A A
C k r
0
= (4)
where
S m
k k k
C
1 1 1
0
+ =
(5)
k
0
: overall rate constant
Limiting cases:
(i)
S m
k k
C
>> (rapid mass transfer), then:
S
k k ~
0
and
A
S
AS
C C ~
i.e. overall process is reaction rate controlled.
(ii)
C
m S
k k >> (rapid reaction), then:
C
m
k k ~
0
and 0 ~
S
AS
C
i.e. overall process is diffusion controlled.
 Secondorder reaction:
Using similar procedure to above, but equation (1) is replaced with:
2
) (
S
A S
S
A
S S
C k r = (6)
gives
A m A
C k r
C
) ) 1 ( (
2 / 1 2
= k k (7)
where
A S
m
C k
k
C
2
1+ = k
i.e. neither 1
st
nor 2
nd
order concentration dependence.
Limiting cases:
(i)
S m
k k
C
>>
2
~
A S A
C k r
(ii)
C
m S
k k >>
A m A
C k r
C
~
Page 8 of 90
For complex reactions, analytical solution is not usually possible.
 Mass transfer can thus lead to difficulties in experimentally determining rate coefficients and
orders.
However, we can work under conditions where we have either reaction or diffusion controlled
process. i.e.

S m
k k
C
>>
or

C
m S
k k >> (in this case should reduce T or increase fluid turbulence)
2.1.2 Attaining Values of k
m
Usually correlations in handbooks define the mass transfer coefficient under conditions of equimolar
counterdiffusion, k
0
m
 How is k
0
m
related to k
m
?
(i) Equimolar counterdiffusion (ECD):
B A
N N
` `
= (8)
dz
dy
CD y N N
A
AB A T A
+ =
` `
(9)
But 0 = + =
B A T
N N N
` ` `
} }
=
=
A
S
S
A
y
y
A AB
l
A
A
AB A
dy CD dz N
dz
dy
CD N
0
`
`
) (
S
A A
AB
A
S
y y
l
CD
N =
`
(10)
But
l
CD
k
AB
m
y
0
(11)
ECD for k k
y y
m m
=
0
(Also,
l
D
C
k
k
AB
m
m
y
C
= = )
(ii) For reaction in which total moles are not conserved, e.g.:
bB aA
A B
N
a
b
N = (12)
equimolar counterdiffusion cannot be used.
Page 9 of 90
Substitute (12) into (9) and rearrange:
} }
=
A
S
S
A
y
y
A
A
AB
l
A
y
a
b
dy
CD dz N
) 1 ( 1
0
) (
S
A A m A
S y
y y k N =
where:
A
y
y
f
m
m
y
k
k
0
=
(13a)
)
1
1
ln(
) 1 ( ) 1 (
S
A A
A A
S
A A A A
f
S
S
A
y
y
y y
y
o
o
o o
+
+
+ +
=
(13b)
a
a b
A
) (
= o (13c)
see Reaction Engineering I

A
f
y
is referred to as film factor
 for the general reaction:
... ... + + + + sS rR bB aA
Equations (13a) and (13b) are applicable but
a
b a s r
A
...) ( ...) ( + + + +
= o
 for 1 , 0
A
f A
y o and
0
y y
m m
k k >>
 common method for predicting k
0
m
is through the use of the j
D
factor.
 j
D
factor:
3
2 0
Sc
G
M k
j
m m
D
(14)
M
m
: average molecular mass (
mol
g
)
G: mass flux (
s m
g
2
)
Sc: Schmidt number
D
f
=
where : viscosity
f
: fluid density
D: molecular diffusivity
k
0
m
can be taken as
0
y
m
k
or
0
f
m
k
, as long as it is remembered that:
A P A P A y
f m f m f m m
P k y P k y k k = = =
0
(
A
f
P
is referred to as pressure film factor)
Page 10 of 90
 in addition to (14), a second relationship for j
D
is available from charts or correlations (see e.g.
Froment & Bischoff, Chemical Reactor Analysis & Design, 2
nd
edn, p.129)
e.g. for flow in a bed packed with spherical particles and c
b
= 0.37:
j
D
= 1.66 Re
0.51
for Re < 190
j
D
= 0.983 Re
0.41
for Re > 190
p
Gd
= Re
Thus, given (14) and a suitable correlation, we can solve for k
0
m
and thus
y
m
k
if
A
f
y
is given (see
section 2.1.3)
 Note: similar correlations also exist for the heat transfer coefficient, h
f
, i.e.:
3
2
Pr
G c
h
j
p
f
H
(15)
(Re) f j
H
=
where Pr: Prandtl number
p
C
=
C
p
: fluid head capacity (e.g. mean value)
: fluid thermal conductivity
2.1.3 Concentration (Partial Pressure) Differences across the External Film
 If AC
A
or AP
A
~ 0 (i.e. Ay
A
~ 0), the mass transfer will be very fast and r
A
can be expressed as a
function of bulk C
A
(or P
A
) directly. E.g.:
A S
S
A A
C k r r
S
= = (since
S
A A
S
C C = )
 Thus it is useful to have some estimate of AC
A
or AP
A
 Well use a different definition of r
A
, i.e. in terms of catalyst mass instead of surface area:
r
A
:
s kg
mol
cat
=
such that, for example:
) (
' S
A A m m A
S C
C C a k r =
where
s m
m
k
p
f
m
C
2
3
: (as before)
cat
p
m
kg
m
a
C
2
:
(see equations (16) and (17) in lecture notes)
(Why redefine r
A
in this way?)
 Method:
y
m m
A
A
k a
r
y
= A
'
Therefore, given a
m
and r
A
, Ay
A
can be calculated if k
my
is known.
Page 11 of 90
A
f
o
m
my
y
k
k = (13a & 13b)
However,
A
f
y
requires
S
A
S
y which is not yet known!
Thus, use an iterative procedure, starting with an initial guess for
S
A
S
y .
A good initial guess is for
A
S
A
y y
S
~ , i.e. Ay
A
= 0, which from equation (13b) gives (using
LHopitals rule):
A A f
y y
A
o + =1
Having attained first estimate of Ay
A
, and thus
S
A
S
y
, the above procedure is repeated until successful
estimates of Ay
A
become negligibly different.
Usually, under practical operating conditions, Ay
A
is negligible, but AT across the external film may
be significant.
2.1.4 Temperature Differences across the External Film
 Energy balance at steadystate:
) ( ) (
'
T T a h H r
S
S m f r A
= A (17)
where:
AH
r
: J/mol
h
f
: J/(m
s
2
s K)
But
A m m A
y a k r
y
A =
'
(18)
) 18 (
) 17 (
(i.e. eliminate r
A
), and substitute for k
my
and h
m
using j
D
and j
H
(eqn 14 & 15) respectively:
A
f
A
p m
r
H
D S
S
y
y
c M
H
Sc j
j
T T T
A A
= A = ) )( ]
Pr
[ ( ) (
3
2
(19)
(AT increases with Ay
A
, i.e. when mass transfer resistance are high)
 For gases flowing in packed beds, the values of the various groups are such that:
A
f
A
p m
r
y
y
c M
H
T
A A
A ) ( 7 . 0 ~ (20)
c
p
: J/(kg K)
M
m
: kg/mol
AH
r
: J/mol
Page 12 of 90
Furthermore, AT is maximum when 0 =
S
A
S
y
(for irreversible reaction, or
m Equilibriu S
A
S
A
y y =
for reversible reaction)
Such that:
) 1 ln(
A A
A A
f
A A
y
y
y
y y
A
o
o
+
=
= A
Eqn (20) then gives:
A
A A
p m
r
y
c M
H
T
o
o ) 1 ln(
) ( 7 . 0 ~
max
+ A
A (21)
2.1.5 Mass Transfer on Metallic Surfaces
 For packed beds, it has been shown that AC variations are small and usually negligible
But mass transfer may be significant when the catalyst is a metallic surface, e.g.:
(1) Catalyst monolith/honeycomb (e.g. catalytic afterburner in automobiles)
(2) Wire gauzes (e.g. oxidation of ammonia)
 Advantages of these units:
(1) low AP
(2) particulate matter in feed will not clog up bed
(difference between monolith and honeycomb reactors?)
 See Fogler p.575, 577 for references on correlations for
C
m
k
for monoliths and wire gauzes.
Page 13 of 90
Monolith supports:
Example
1
st
order oxidation of CO in a catalytic converter. Calculate:
(1) Exit molar flow rate of CO, n
A
, given n
A0
, k
S
,
C
m
k
, and a
v
(
3
2
r
S
m
m
).
(2) What is the corresponding equation for very rapid reaction kinetics?
Solution:
Assume PFR, constant volumetric flow rate (v
0
):
(i)
V A A m A
A
a C C k r
dV
dn
S C
) ( = = (*)
0 0
,
) (
v
n
C
v
n
C
C C k C k
S
S
S C S
A
A
A
A
A A m A S
= =
=
Thus, show:
V
a k
A A
e n n
t
0
0
= where t = V/v
0
(ii) 0 ~ ,
S
A m S
S C
C k k >>
Equation (*) can be directly integrated to give:
V
C
m
a k
A A
e n n
t
=
0
Page 14 of 90
2.2 Intraparticle Gradient Effects
2.2.1 Catalyst Internal Structure
 Reaction rate catalyst surface area
 Areas range from 10200 m
2
/g (most towards higher end); activated carbon ~ 800 m
2
/g
(c.f. sand, 0.01 m
2
/g)
 High areas through a highly porous structure, i.e. high surface area to volume ratio.
 Pore size not uniform, i.e. a pore size distribution (PSD) exists.
 PSD measured by (nitrogen) porosimetry. Suitable for measuring pore sizes in the range of
130 nm.
(How does porosimetry work?)
 Pore size often classified as:
(1) micropores: d
pore
< 0.3 nm
(2) mesopores: 0.3nm < d
pore
< 20 nm
(3) macropores: d
pore
> 20 nm
 Often use a mean pore size in calculations
For some catalysts, can have a bimodal distribution of pore sizes (e.g. zeolite catalysts).
What are zeolites?
What is a bimodal distribution?
What is a bidisperse structure?
 For nonzeolitic catalysts, active metal dispersed and supported within a macroporous
support matrix, such as silica or alumina.
Thus unimodal PSD, but can be rather broad.
2.2.2 Pore Diffusion
 For gas diffusion through a single cylindrical pore, the ratio of d
pore
to the mean free path of the
gas () will determine whether or not the pore well affects the diffusion behaviour.
(What is meant by the mean free path of the gas?)
 3 different situations:
(i) d
pore
>>
 molecular diffusion dominates, i.e. Fickian diffusion
 e.g. gases at high pressure; liquids
d
pore
. . . .
. . . .
. . . .
. .
Page 15 of 90
(ii) d
pore
<< , and d
molecule
< d
pore
 molecular interactions with the pore wall dominate
 diffusion described by Knudsens law
 e.g. gases at low P, but not liquids (why?)
(iii) d
pore
<< , and d
molecule
~ d
pore
 complex interaction of diffusing molecules with the forcefields of molecules making up the
wall
 referred to as configurational diffusion
 very difficult to predict
 e.g. very large hydrocarbon molecules (e.g. petroleum desulpurisation); pores of very small size
(e.g. diffusion within zeolite crystals, and through biological cell walls)
(see Froment & Bischoff, Chemical Reactor Analysis & Design, p.143)
D
i
f
f
u
s
i
o
n
c
o
e
f
f
i
c
i
e
n
t
(
c
m
2
/
s
)
d
pore
/2 , (nm)
Page 16 of 90
 Correlations for Diffusion Coefficients
 correlations for binary molecular diffusion can be attained from Handbooks.
For gases,
P
T
D
k i
m
2
3
,
 usually need an effective binary diffusion coefficient, i.e. diffusion coefficient for the key
component through a mixture of the other components,
m i
m
D
,
.
i.e.
dz
dy
D C N y N
i
m
N
k
k i i
m i
C
,
1
=
=
N
c
: number of components
 given the StefanMaxwell equations for diffusion, we can calculate
m i
m
D
,
from actual binary
diffusivity data,
k i
m
D
,
:
=
=
=
C
C
k i
m i
N
k i
k
i
N
i k i
k
i k
m
m
v
v
y
v
v
y y
D
D
1
1
) (
1
1
,
,
where v is the stoichiometric coefficient.
 Knudsen diffusion coefficient, D
k
, can be calculated from the kinetic theory of gases as:
pore
M
k
d
M
T
D
i
i


.

\

2
1
i.e. ) (P f D
i
k
=
But, as P is increased, the transport regime can switch from Knudsen to molecular diffusion.
 as mentioned earlier, configurational (or micropore) diffusion coefficient is difficult to predict
measurement needed.
 for nonzeolite catalysts (and for the binder phase of zeolite catalysts) molecular and Knudsen
diffusion dominate. The pore diffusion coefficient, D
p
, will thus be a function of D
m
and D
k
.
(D
p
: pore diffusion coefficient for a single pore)
Treybal (1981) suggests that for:
(1) d
pore
/ > 20 , i.e. molecular diffusion controlling
D
p
= D
m
(2) d
pore
/ < 0.2 , i.e. Knudsen diffusion controlling
D
p
= D
k
For intermediate values, both diffusion types are important. Can use Bosanquet formula to
approximate D
p
for this case:
m k p
D D D
1 1 1
+ =
(c.f. resistance in parallel)
Page 17 of 90
 given D
p
, how do we calculate an effective diffusion coefficient, D
e
, which account for the
complex pore structure of the catalyst pellet?
An approximation of D
e
(but often adequate for design purposes) is given by:
p
p p
e
D
D
t
c
= (22)
where
c
p
: intraparticle void fraction (
3
3
p
p
m
m
V
)
t
p
: tortuosity factor
Basis for equation (22):
Compare diffusion in a single pore and diffusion in a porous pellet:
versus
 crosssectional area available for diffusion = A c
p
lower N
A
 tortuous molecule path, and changing pore crosssectional area due to constrictions. So
dz
dC
A
is
reduced
Thus:
dz
dC
D
N
A
p
p p
A
t
c
=
[ Note: t
p
= tortuosity/(constriction factor)
where tortuosity = (actual diffusion path length)/(shortest length) ]
Given D
e
, we can now consider combined diffusion and reaction within a catalyst pellet.
Unlike reaction at a surface, diffusion and reaction take place simultaneously for this case rather than
consecutively.
C
A1
C
A2
z
dz
dC
D N
A
p A
=
A
linear molecule
path on average
Page 18 of 90
2.2.3 Reaction and Diffusion within a Catalyst Pellet
 Consider the concentration profiles within a porous catalyst pellet:
 Chemist measures rate under conditions where external and internal mass transfer resistances are
negligible, say r
A
*
.
How?
 When mass transfer is important:
C
A
> C
As
Thus we cannot use bulk concentrations to calculate actual (observed) reaction rate, r
A
.
We need to relate r
A
to r
A
*
. This is done through the effectiveness factor, q:
position
external
film
Concentration
negligible external film resistance
significant external film resistance
external
film
C
A
C
S
As
C
As
(catalyst)
r
p
r
0 (centre/central axis of pellet)
Page 19 of 90
*
A
A
r
r
= q (<1 for isothermal or endothermic reactions)
 q is useful in design calculations. But for rigorous calculations, particularly for complex reaction
kinetics and nonisothermal operation, it is better to solve the simultaneous equations governing
diffusion and reaction (see section 3).
 have previously shown that for a packed bed, external film mass transfer resistances are small.
Thus we can assume the situation depicted by the solid line in the previous graph.
In other words, r
A
*
is the reaction rate which would be measured if all of the pellet saw a
concentration of C
S
As
:
S
A
S
A A A
S S S
r C k r = = ] [
*
and
S
A
A
S
r
r
= q
(We will later consider the case when both external and intraparticle gradients are important.)
(Pseudo) First Order Reaction (A Product)
 Consider material balance through an incremental section of a catalyst slab of crosssectional
area a:
C
A
= C
S
As
C
S
As
C
As
r
p
r
0
N
A
==>
(mol/m
2
p
s)
r
r +Ar
Page 20 of 90
IN OUT = CONSUMPTION
r a r a N a N
S
A
r
A
r r
A
A =
A +
) ( ) (
divided by (aAr) and let Ar 0:
S S
A v A
A
C k r
dr
dN
= =
where
r
As
: mol/(m
3
p
s)
C
As
: mol/m
3
f
 For no convective flow in the pellet:
dr
dC
D N
S
A
A
e A
=
S
S
A
A v
A
e
C k
dr
C d
D =
2
2
(for D
eA
being constant with r)
Integrating this differential equation with the boundary conditions:
r = r
p
C
As
= C
S
As
(=C
A
)
r = 0
0 =
dr
dC
S
A
gives:


.

\



.

\

=
A
A
S
S
e
v
p
e
v
S
A
A
D
k
r
D
k
r
C
C
cosh
cosh
(24)
Note:
A
e
v
p slab
D
k
r 
(Thiele Modulus), i.e.
Page 21 of 90
 We can derive C
As
profile for a spherical pellet in a similar way:
i.e. a = 4tr
2
IN OUT = CONSUMPTION
r r r N r N r
S
A
r
A
r r
A
A =
A +
2 2 2
4 ) ( 4 ) ( 4 t t t
divided by 4tr
2
and let Ar 0:
S
A v A
C k N r
dr
d
r
= ) (
1
2
2
(25)
Again, using
dr
dC
D N
S
A
A
e A
=
and the same boundary conditions as for a slab (r
p
= pellet
radius for this case) gives (after integration):


.

\



.

\

=
A
A
S
S
e
v
p
e
v
p
S
A
A
D
k
r
D
k
r
r
r
C
C
sinh
sinh
(26)
Note:
A
e
v
p sphere
D
k
r 
 Likewise for a cylindrical pellet:
S
A v A
C k rN
dr
d
r
= ) (
1
(27)
Boundary conditions:
r = r
p
C
As
= C
S
As
r = 0
0 =
dr
dC
S
A
Integrating gives:


.

\



.

\

=
A
A
S
S
e
v
p
e
v
S
A
A
D
k
r I
D
k
r I
C
C
0
0
(28)
I
0
: Bessel function
Note:
A
e
v
p cyl
D
k
r 
 C
A
/C
As
vs r/r
p
profiles of cylindrical and spherical pellets are similar to slab pellets
r
p
Page 22 of 90
 Comparing equations (23), (25), and (27), we can write the general pellet mass balance as:
S
A A
m
m
r N r
dr
d
r
= ) (
1
(29)
where
m = 0 for slab
m = 1 for cylinder
m = 2 for sphere
and
dr
dC
D N
S
A
A
e A
=
Boundary conditions:
r = r
p
C
As
= C
S
As
r = 0
0 =
dr
dC
S
A
where r
p
= characteristic halflength of pellet
 Given C
As
= f(r) for a firstorder reaction, we can calculate the average reaction rate ( observed
rate):
p
V
A
p
A A
dV r
V
r r
p
S S }
= =
0
1
(30)
Given r
A
, we are now in a position to calculate the effectiveness factor, q.
Effectiveness Factor: FirstOrder Reactions
 Recall:
q
= (observed reaction rate)/(reaction rate at pellet surface conditions)
S
A
A
S
r
r
= q
(31)
For isothermal (and endothermic) reactions, r
S
As
represents the maximum reaction rate, since
C
S
As
> C
As
, or in other words:
1
) (
s
= > =
q
A A v
S
A v
S
A
r C k C k r
S S S
Also, if there is very high resistances within catalyst, there is negligible penetration of reactant
into the pellet such that:
0 , 0 , 0 = = = q
S S
A A
r C
1 0 s s q
 If q can be calculated, equation (31) can be used to determine r
A
, since r
S
As
is readily calculated,
i.e.
S
A A
S
r r q =
where
A v
S
A v
S
A
C k C k r
S S
~ =
for negligible external film mass transfer resistances.
Page 23 of 90
How do we do this?
 For a slab:
 Given C
As
from equation (24) and r
As
= k
v
C
As
, substitute into equation (30) and integrate to
calculate ) (
A A
r r
S
=
 Given r
A
, use equation (31) to derive an expression for q.
 Solution:
slab
slab


q
) tanh(
= (32)
where
A
e
v
p slab
D
k
r = 
(33)
Note:
(34)
 For a sphere (using a similar procedure for slab, but use equation (26) for C
As
):


.

\

=
sphere sphere sphere
  
q
1
tanh
1 3
(35)
(
sphere
 as for
slab
 but r
p
= pellet radius)
Note:
>
sphere
sphere
sphere
e i

q 
q 
3
~ ), 9 . . (
1 , 0
(36)
 Similarly, for a cylinder:
cyl cyl
cyl
I
I
 

q
1
) 2 (
) 2 (
0
1
=
(37)
Note:
(38)
>
cyl
cyl
cyl
e i

q 
q 
2
~ ), 5 . . (
1 , 0
>
slab
slab
slab
e i

q 
q 
1
~ ), 3 . . (
1 , 0
Page 24 of 90
q versus  for 1
st
Order Reaction
1: slab
2: cylinder
3: sphere
 q versus  equations are rather complex for spheres and cylinders.
But we can see from plots that the trends are very similar but a shift along the xaxis.
We can, in fact, redefine Thiele modulus such that for any pellet geometry, q versus 
approximately coincide, i.e.
Curves for sphere and cylinder coincide with slab curve, such that the relatively simple
expression:


q
tanh
=
can be used, where  is a redefined (general) Thiele modulus,
This is achieved for:
A
e
v
p
p
D
k
A
V
= 
(39)
where:
V
p
: particle volume
A
p
: external surface area of particle
i.e.
slab: V
p
/A
p
= r
p
( = 
slab
)
sphere: V
p
/A
p
= r
p
/ 3
cylinder: V
p
/A
p
= r
p
/ 2
Page 25 of 90
 Given , equation (32) is applicable to all pellet geometries. Maximum discrepancy between
curves ~ 15%.
What are the limits of q for 0 and  ?
 Example 3
Reaction rate measured in lab under conditions where intraparticle and externalfilm resistances
to mass transfer are negligible, i.e.
r
As
= 1.5 C
S
As
(C
A
= C
S
As
for this case)
What is the actual (observed) reaction rate in an industrial reactor where intraparticle gradients
are important?
Ans: r
A
= q 1.5 C
A
where


q
tanh
=
(1
st
order reaction)
To calculate , we need V
p
/A
p
, k
v
(=1.5), and D
eA
.
What if the reaction is not 1
st
order?
Effectiveness Factor: General Order Reactions
 For general order and reversible reactions, we can further generalise the Thiele modulus as:
2
1
*
2

.

\

=
}
S
S
A
S
A
S S A
S
C
C
A A e
S
A
p
p
dC r D
r
A
V

(40)
where C
*
As
is the equilibrium concentration of the limiting reactant (=0 for an irreversible
reaction).
(See Froment & Bischoff, p.160162)
 Equation (40) accounts for D
eA
variation with C
As
.
 Equation (40) assumes diffusional resistances are high such that we are in the q ~ 1/ region. If
not, C
*
As
in equation (40) needs to be calculated from:
p
C
C C
C
A A e
A e
r
dC r D
dC D
S
S
A
S
A A
S
A
S A
S A
=

.

\

}
}
* '
*
'
2
(+)
Page 26 of 90
 Example:
n
th
order irreversible reaction of A in a spherical catalyst pellet, for which D
eA
~ constant (assume
strong diffusion limitations)
Ans:
S
n
A v A
p
p
p
C k r
r
A
V
S
=
=
3
2
1
0
1
1
1
2 3
) (
+


.

\

(
+
=
S
S
A
S A
S
C
n
A e v
n S
A v p
C
n
D k
C k r

2
1
1
) (
2
1
3


.

\

+
=
A
S
e
n S
A v p
D
C k
n
r

(for n > 1)
Again, C
S
As
~ C
A
for typical packed bed reactors.
Note: for a non1
st
order reaction, q = f(C
A
) and will therefore vary with axial position in a
tubular reactor.
Criteria for Intraparticle Diffusional Limitations
 If reaction kinetics are known, q can be calculated; q < 1 indicates diffusional limitations.
However, in experiments where we need to calculate kinetics (e.g. rate constants), we need to
ensure that we are working in a region where diffusional limitations are negligible so as not to
falsify the kinetic data.
 WeiszPrater criteria:
Rearrange equation (39):
v e
p
p
k D
V
A
A
=


.

\

2
2

S
A v
S
A A
S S
C k r r q q = = (1
st
order reaction)
Eliminate k
v
and rearrange:
u = =


.

\

2
2
 q
p
p
S
A e
A
A
V
C D
r
S A
(41)
(C
S
As
~ C
A
under typical operating conditions)
L.H.S. of (41) is measurable. Then for:
(i) negligible diffusional limitations:
1
1 ~ , 1
<< u
<< q 
Page 27 of 90
(ii) considerable diffusional limitations:
1
1
~ , 1
>> u
>>

q 
 The above method can be generalised to any reaction scheme using the appropriate form of the
Thiele modulus (see eqn.(40)); see also Froment & Bischoff, p.169)
2.2.4 Temperature Gradients within a Catalyst Pellet
 We can calculate temperature gradients within a catalyst pellet by considering simultaneously the
intraparticle mass and energy balances:
Eqn.(25) (for a spherical pellet):
S
S
A
A
A
e
r
dr
dC
r
dr
d
D
r
= ) (
1
2
2
Similarly for energy balance, given an effective thermal conductivity,
e
:
r A
S
e
H r
dr
dT
r
dr
d
r
S
A = ) (
1
2
2
(42)
Exercise: Eliminate r
As
between (25) & (42) and integrate twice to obtain:
) ( ) (
S
A A
e
e r S
S S S
S S
A
C C
D H
T T T
A
= = A
 AT
S
is maximum when C
As
= 0 for an irreversible reaction (or C
*
As
for an equilibrium
(reversible) reaction).
i.e.
S
A
e
e r
S
S
A
C
D H
T
A
= A
max
(43)
 It can actually be shown that eqn.(43) is applicable to all pellet geometries, i.e. same eqn. Is
derived.
 For many industrial applications:
1 . 0
max
<
A
S
S
S
T
T
i.e. AT
S
is small, but AT(externalfilm) can be large (other way around to AC
A
). Exceptions
include highly exothermic reactions such as some oxidation and hydrogenation reactions.
 Effect of AT
S
on q is complex, e.g. will influence D
eA
as well as k
v
. For highly exothermic
reactions, q can exceed 1. (Why?)
Page 28 of 90
Example: q for 1
st
order reaction in a nonisothermal pellet:
 Use equations (25) & (42) and
S
S
S
A
RT
E
A
C e A r
=
0
 Put in dimensionless form by defining:
p
S
S
S
S
A
A
r
r
r
T
T
T
C
C
C
S
S
= = = ,
Gives:
)
1
1 (
2
2
2
)
1
1 (
2
2
2
) ' (
) ' (
T
T
e C
r d
T d
e C
r d
C d
=
=
 

where:
A
e
p
D
e A r

=
0
2
2
) ' (
S
S
RT
E
=
(Arrhenius number)
S
S e
S
A e r
S
S
S
T
C D H
T
T
S A

A
=
A
=
max
Solution is a function of  , , and  only. (See Weisz & Hicks (1962) diagram)
 Liu (1969) shows:
5
'
1
~


q e
(See Froment and Bischoff, p.184)
Page 29 of 90
(ref. Weisz & Hicks (1962))
 = 0 isothermal
 < 0 endothermic
 > 0 exothermic

Page 30 of 90
2.3 Combined Interfacial (External) and Intraparticle (Internal)
Resistances
 In solution of intraparticle diffusion equation, C
S
As
was assumed known (i.e. = C
A
) and constant.
When externalfilm resistances are important, the boundary conditions for the solution of the
intraparticle diffusion equation become:
p
S
A S C
r
A
e
S
A A m p
dr
dC
D C C k r r = = ) (
0
0
dr
dC
r
S
A
=
(as before)
 E.g. for slab pallet with a 1
st
order reaction, solution of eqn.(23) with the above boundary
conditions gives:




sinh cosh
) cosh(
C
A
S
m p
e
p
A
A
k r
D
r
r
C
C
+
=
 Can then define a global effectiveness factor, q
G
, as:
q
G
= (rate observed) / (rate at bulk fluid concentrations)
] [
A A
A
C r
r
S
=
which gives:
m G
Bi
2
1 1 
q q
+ =
(44)
where
A
C
e
p m
m
D
r k
Bi =
(Biot number for mass transfer , or Sherwood number)
i.e. For Bi
m
>> 1, q
G
= q.
 In the region of strong intraparticle diffusional limitations,

q 
1
, =
:
m G
Bi
2
1 

q
+ =
(45)
 Finally, for other reaction schemes (non1
st
order_ and pellet geometries, Aris (1965) has shown
that eqn.(45) is again applicable. But  is calculated using generalised Thiele modulus (eqn.(40))
for which C
As
is replaced by C
A
.
Page 31 of 90
3 Fixed Bed Catalytic Reactor (FBCR) Design
 We will now consider mathematical models describing FBCRs
 We will start with simple homogeneous models, then models accounting for interfacial and
intraparticle gradients through the use of:
(i) an effectiveness factor, or
(ii) actual pellet phase mass and energy balances
 We will principally consider 1D models, i.e. no radial gradients in C or T.
 We will also consider a single reaction, in which the consumption of specie i is denoted by r
i
.
(extension of models to multiple reactions will be demonstrated)
 The key reactant will be denoted by A
3.1 PseudoHomogeneous PFR and Axially Dispersed PFR Models
3.1.1 PFR Model
 Simplest FBCR model (from Reaction Engineering I):
b A
A
i
b i i
i
r
v
v
r r
dV
dn
' '
= = =
(1)
 Noting that: n
i
= u a C
i
, dV = a dz:
b A
A
i
b i i
r
v
v
r uC
dz
d
' '
) ( = =
(2)
 If u constant, we need a model for velocity distribution, i.e. an approximation of the
momentum equation.
E.g. Ergun equation:
2
2 1
u E u E
dz
dP
i
=
(3)
where:
3
2
3 2
2
1
) 1 ( 8 . 1
) 1 ( 180
b p
m g b
b p
b
d
M
E
d
E
c
c
c
c
=
(For laminar flow, the term containing E
2
can be neglected; for highly turbulent flow, the term
containing E
1
can be neglected.)
Page 32 of 90
 For a perfect gas:
g
i
i
RT
P
C = =
(4)
 For nonisothermal operation, energy balance needed to describe Tz variation.
From Reaction Engineering I (basis: J/(m
3
s)):
0 ) (
'
= A +
v r A b p
i
i
a Q H r C n
dV
dT
i
(5)
where:
Q = U(T
C
T) (J/(m
2
s))
a
v
= (surface area) / (reactor volume) (m
1
)
[U: overall heat transfer coefficient, J/(m
2
s K)
T
C
: temperature of cooling (heating) fluid, K ]
0 ) (
'
= A +
v r A b p
i
i
a Q H r C C U
dz
dT
i
(6)
Do we need to solve eqn.(1) for all reaction species?
 For no separation of reaction species due to, say, different rates of axial dispersion or
intraparticle diffusion (see later notes), we can relate C
i
(i A) to C
A
from reaction stoichiometry:
(n
Ao
n
A
) = mol A reacted
) (
0 0
A A
A
i
i i
n n
v
v
n n + =
i.e.
) (
0 0
0 0 A A
A
i
i i
uC C u
v
v
C u C u + =
(7)
Other way of relating C
i
of all components and thus reducing material balance to one equation is
the use of conversion or extent of reaction!
 Boundary conditions:
Equations (2), (3), and (6) are 1
st
order ODEs. Thus each requires 1 boundary condition for the
dependent variable. Usually we specify conditions at the reactor entrance. Thus:
=
=
= = =
0
0
1
) .. 1 ( : 0
0
T T
P P
N i C C z
C i i
(8)
 Check on degrees of freedom:
Unknowns: N
C
C, u, P, T
i.e. N
C
+ 3
Equations: (2), (3), (4), (6), (7)
(N
C
1)
i.e. N
C
+ 3
Therefore, solution possible.
Page 33 of 90
3.1.2 Axially Dispersed PFR Model
 Revision from Reaction Engineering I:
Material balance on incremental section:
where D
Z
= axial dispersion coefficient (m
2
/s)
D
Z
0 for PFR
D
Z
for CSTR
0 ) ( ) (
'
=
b A
A
i i
Z i
r
v
v
dz
dC
D
dz
d
uC
dz
d
i
(9)
Similarly for the energy balance, we can have an effective thermal dispersion coefficient in the
axial direction, k
Z
, to give:
0 ) ( ) (
'
= A +
c
c
c
c
c
c
v r A b Z p
i
i
a Q H r
z
T
k
z
C C u
z
T
i
(10)
 Equations (3) & (4) are again applicable.
 Eqn.(9) is only applicable if the dispersion coefficients for all species are the same, i.e. D
Zi
= D
Z
.
(Why?)
Otherwise, we need to solve eqn.(9) for all i.
 Boundary conditions:
Now we have 2
nd
order ODEs for C & T (equations (9) & (10)). So we need 2 boundary
conditions for each equation. These are chosen at z = 0 and z = L using the Danckwerts boundary
conditions.
 Danckwerts boundary conditions account for dispersion in the entrance/exit regions of the bed:
z = 0:
0
0 0
=
=
=
Z
i
Z
Z
i i
dz
dC
D uC uC
(I)
i.e.
0 = Z
i
C gives actual boundary condition for C
i
at z = 0.
uC
i o
z = 0
.
) (
dz
dC
D uC
i
Zi i
+
z z + Az
) (
dz
dC
D uC
i
Zi i
+
z a r
b i
A '
Page 34 of 90
z = L:
ie i
L Z
i
Z
L Z
i
uC
dz
dC
D uC =
=
=
(II)
But
ie i
C C > (otherwise physically meaningless!)
0 >
=L Z
i
dz
dC
If
0 <
=L Z
i
dz
dC
, from (II):
ie i
C C < ! which is an inconsistency.
Thus, the only sensible (feasible) solution for (II) is
0 =
=L Z
i
dz
dC
In summary, Danckwerts boundary conditions are given by:
= =
= =
=
=
=
0
) ( 0
0
0 0
L Z
i
Z
i
Z i
Z
i
dz
dC
L z
dz
dC
D C C u z
(11)
 Danckwerts boundary conditions can be derived for energy balance in a similar way, i.e.:
=
c
c
=
c
c
= =
=
=
=
0
) ( 0
0
0
L Z
Z
Z
Z
i
p i
z
T
L z
z
T
k T T C C u z
i
(12)
 Boundary condition for eqn.(3) is the same as before, i.e. P = P
0
at z = 0.
 Eqn.(11), for z = 0, again only needs to be specified for (N
C
1) components, since C
i
for N
c
th
component is given by eqn.(4) if P
0
and T
0
are specified.
But, for z = L condition, we need to be a bit more careful!
 If P
e
is held constant, then
g e
will be known. Thus (N
C
1) equations need to be specified for
L Z
i
dz
dC
=
.
uC
i e
z = L
.
Page 35 of 90
 If the outlet flow through a valve (e.g. to control gas velocity):
) ( P f u
L Z
A =
=
(valve equation)
This sets
L Z
dz
dP
=
from eqn.(3),
L Z
P
=
, and again (N
C
1) equations need to be specified for
L Z
i
dz
dC
=
.
 Degrees of freedom analysis as for model in section 3.1.1.
 When is axial dispersion important?
Young and Finlayson (1973) show that for:
0
'
) (
A
p b A
Z
p
uC
d r
Pe
D
d u
>> =
Axial dispersion can be neglected.
Note: Peclet number in this case is defined in terms of d
p
rather than L.
As a rule of thumb, for flow velocities used in industrial practice, the effect of axial dispersion
(head and mass) on conversion is negligible when:
L > 50 d
p
P
e
u 
z =L
P
z = L
Page 36 of 90
3.2 Heterogeneous Models
3.2.1 Use of Effectiveness Factor
 If q can be readily calculated, we can use this in our material and energy balances.
i.e. replace r
A
in equations (2) & (6), or (9) & (10) with:
q r
A
or q
G
r
A
to give actual (observed) reaction rate.
(remember: q
(G)
can be a function of z)
 E.g. PFR model:
(i) If q
G
is given, the equations given in section 3.1.1 are applicable. But r
A
needs to be replaced
with q
G
r
A
.
(ii) If q is given, then if we neglect external film mass & heat transfer resistances, the equations
given in section 3.1.1 are applicable. But r
A
needs to be replaced with q r
A
.
Otherwise:
(a) If we allow for external film heat transfer resistance only, the bed energy balance (eqn.(6)) is
now given by:
0 ) ( ) (
'
= +
S
S m f v p
i
i
T T a h a Q C C u
dz
dT
i
(13)
Another variable, T
S
S
, is now introduced but can be found from pellet phase energy balance:
r b A
S
S m f
H r T T a h A = q
' '
) ( (14)
(What are the units of a
m
?)
(b) If we allow for both external film heat and mass transfer resistance, in addition to equations
(13) & (14), we need to replace material balance with:
0 ) ( ) (
'
=
S C
i i m m i
C C a k uC
dz
d
(15)
and:
b A i i m m
r C C a k
S C
q
' '
) ( = (16)
 The boundary conditions given in section 3.1.1 are again applicable for this case. Analogous
equations can be written in the presence of mass and thermal dispersion, i.e. equations (14) &
(16).
Page 37 of 90
3.2.2 Use of Intraparticle Diffusion Equations
 We solve for intraparticle diffusion equations when q cannot be determined readily, e.g. complex
reaction kinetics and nonisothermal operation.
i.e. we solve diffusion equations for mass and heat in a catalyst pellet simultaneously with bed
mass and heat balance. The former were given in sections (2.2.3) & (2.2.4) as (spherical pellet):
b A
A
i
i e
S
S i
r
v
v
dr
dC
r
dr
d
r
D
' 2
2
) ( =
(17)
r b A
S e
H r
dr
dT
r
dr
d
r
S
A =
' 2
2
) (
(18)
Equations (17) & (18) are solved at each z so as to give, for example,
S
i
S
C and T
S
S
, so as to be
able to calculate mass and heat transfer rates from bed to pellet at each z.
 In most rigorous form, equations (13) & (15) are applicable for this case (or the analogous
equations in the presence of mass and thermal dispersion), as well as corresponding boundary
conditions.
Now need boundary conditions for (17) & (18). They are 2
nd
order ODEs in
S
i
C and T
S
. So 2
boundary conditions are needed for each:
r = 0: (symmetry condition)
0
0
0
0
=
=
=
=
r
S
r
i
dr
dT
dr
dC
S
r = r
p
:
p
p
S
i S C
r r
S
e
S
S f
r r
i
e
S
i i m
dr
dT
T T h
dr
dC
D C C k
=
=
=
=
) (
) (
Exercise: carry out a degrees of freedom analysis for the above model.
Page 38 of 90
3.3 2D Models
 2D models account for radial variations in C & T (& T
S
) in addition to axial variations.
These usually arise due to a combination of high AH
r
and nonadiabatic operation through
heating/cooling of the tube (reactor) walls.
Radial velocity gradients will again be associated with mass and thermal dispersions, which can
be characterised by
y
z
D and
y
z
k , where y denotes the radial dimension in the tube.
Radial dispersion will be much more significant than axial dispersion. Thus in 2D models, the
latter is usually neglected.
How do we derive mass and energy balances for this case?
 Need to consider an incremental annulus within the bed, e.g.:
 Thus, for example, by neglecting interfacial and intraparticle gradients we can derive (pseudo
homogeneous model):
0 )
1
( ) (
'
2
2
=
c
c
+
c
c
c
c
b A
A
i i i
z i
S y
r
v
v
y
C
y y
C
D uC
z
(19)
0 )
1
( ) (
'
2
2
= A +
c
c
+
c
c
c
c
v r A b z
i
p i
a Q H r
y
T
y y
T
k C C u
z
T
S y i
(20)
The equations are now 2
nd
order with respect to C
i
& T in the ydomain. Thus 2 additional
boundary conditions in the ydomain are needed. There are:
y = 0: (centre of tube)
0
0
0
0
=
c
c
=
c
c
=
=
y
y
i
y
T
y
C
Az
Z
i
Z i
dz
dC
D uC
Z Z
i
Z i
dz
dC
D uC
A +
y
i
Zy
dy
dC
D
y y
i
Zy
dy
dC
D
A +
Page 39 of 90
y = y
w
: (tube wall; y
w
= tube radius)
0 =
c
c
=
w
y y
i
y
C
(Why?)
) (
w
y y
w
y y
z
T T
y
T
k
w
w
y
=
c
c
=
=
o
Where o
w
is an effective heat transfer coefficient at the inner surface of the tube, and T
w
the wall
temperature.
o
w
is difficult to predict due to variations in gas flow and bed packing in the vicinity of the wall
compared to the rest of the bed.
 For a 2D heterogeneous model, researchers have shown that for the radial dispersion of heat, it is
best to distinguish between the solid and fluid phases.
Thus, neglecting externalfilm mass transfer resistance:
 eqn.(19) is applicable but r
A
is replaced by q r
A
 eqn.(20) is rewritten as:
0 ) ( )
1
( ) (
'
2
2
= +
c
c
+
c
c
c
c
s m f
f
z
i
p i
T T a h
y
T
y y
T
k C C u
z
T
y i
(21)
 eqn.(14) is rewritten as:
)
1
( ) (
2
2
' '
y
T
y y
T
k H r T T a h
S
z r b A
S
S m f
y
c
c
+
c
c
A = q
(22)
(f & s denote fluid and solid (pellet) phases respectively)
What are the boundary conditions for the above model?)
 Area for background reading:
(i) Multiple steady states in exothermic reactions (see Fogler).
keywords: ignition and extinction temperatures
(ii) Reaction runaway: critical inlet conditions for reaction runaway.
Page 40 of 90
4 Fluidised Bed Reactors
These involve catalyst beds which are not packed rigid but either suspended in fluid (i.e. fluidised
bed reactors), or flowing with the fluid (i.e. transport reactors).
4.1 Overview of Fluidisation Principles
For downward flow in a packed bed, there is no relative movement between particles. AP u for
laminar flow, or AP u
2
for highly turbulent flow.
For upward flow through the bed, AP is the same as downward flow at low flow rates. But when
frictional drag on particles becomes equal to their apparent weight (i.e. actual weight less buoyancy),
the particles become rearranged such that they offer less resistance to the flow, and the bed starts to
expand. As u increases, the process continues until the bed has assumed its loosest stable form of
packing. Particles become freely supported in the fluid, and the bed is said to be fluidised.
Minimum fluidisation velocity, u
mf:
the fluid velocity at the point where fluidisation occurs.
At superficial fluid velocities > u
mf
:
(i) Liquid fluidisation: bed continues to expand with u, and maintains a uniform character, bet
agitation (mixing) of particles increases. i.e. particulate fluidisation.
(ii) Gas fluidisation: gas bubble formation within a continuous phase consisting of fluidised
solids The continuous phase referred to as the dense or emulsion phase. i.e. aggregative
fluidisation.
For aggregative fluidisation, even at high inlet flow rates, the flow in the emulsion phase relative to
the particles remains roughly constant. But bubbling may be more vigorous.
For high flow rates and a deep bed, bubbles can coalesce, and even form slugs of gas which occupy
the entire crosssection of the bed.
 For our fluidised bed reactors, we will be principally concerned with gas fluidisation.
 Fluidised bed behaves like a fluid:
hydrostatic forces are transmitted (why is this useful?), and solid objects float if their densities <
that of the bed.
There is intimate mixing and rapid heat transfer. Thus it is relatively easy to control T.
 the type of fluidisation (see attached figure) depends on particle size and relative density of
the particles (
s
g
).
Geldart classification can be used to give an indication of the type of fluidisation achievable (see
figure below).
 Why fluidisation?
 As mentioned above, we can achieve good control of T (even for highly exothermic reactions)
 Can work with very fine particles, for which q ~ 1
Page 41 of 90
 As catalysts improve, rates of reaction increase, i.e. higher values of k
v
.
But
A
e
v
p
D
k
r
=
3

(spherical pellet)
D
eA
~ constant. Thus as k
v
increases, the only way to keep  small (and thus q close to 1) is to
decrease r
p
.
Page 42 of 90
(Why not use small particles in a packed bed?)
 AP vs. u
o
in a fluidised bed:
where:
u
0
= superficial velocity at bed inlet
u
t
= terminal velocity, i.e. pellets blown out of bed
Note: in fixed bed region:
laminar flow:
0 1
u E
L
P
=
A
) log( ) log(
0
u C P + = A
turbulent flow:
2
0 2
u E
L
P
=
A
) log( 2 ) log(
0
u C P + = A
slope=1
for laminar flow
Page 43 of 90
 Calculation of AP across a fluidised bed:
For a fluidised bed:
Total frictional forces on particles = effective (apparent) weight of particles
F
1
= F
2
g L A A P A P
b b
S
S
m
kg
m
m
m
kg
g s
+ =
= 
_
3 3
3
3
) 1 )( (
2 1
c
g L P P P
g s
= = A ) 1 )( ( ) (
1 2
c (1)
If there is an increase in P
1
, AP at that instant will become higher. But then c increases (bed
expands), or lowresistance gas bypass through bubbling, such that AP remains the same.
 Calculation of u
mf
:
(i) If laminar flow at the point of fluidisation
g u E
L
P
g s mf mf
mf
mf
) )( 1 (
1
c = =
A
1
) )( 1 (
E
g
u
g s mf
mf
c
=
For
3 2
2
1
) 1 ( 180
mf p
mf
d
E
c
c
= :
c
c g d
u
g s
mf
p mf
mf
) (
) 1 ( 180
1
2 3
= (2)
(c
mf
~ 0.4 for a bed of isometric particles)
(ii) If turbulent flow at the point of fluidisation (usually the base for coarse particles)
g u E u E
L
P
g s mf mf mf
mf
mf
) )( 1 (
2
2 1
c = =
A
Thus we can solve for u
mf
explicitly.
However, it is convenient to work in terms of dimensionless groups:
Page 44 of 90
(1)
2
3
) (
p g s g
d g
Ga
(Galileo number)
(2)
p mf g
mf
d u
Re
to give (using the 3 equations above):
2
3 3
Re
75 . 1
Re
) 1 (
180
mf
mf
mf
mf
mf
Ga
c c
c
+
= (3)
 In reality we expect Darcys law and Ergun equation to overestimate AP
mf
. (Why?)
For laminar flow, investigations have shown that it is more accurate to use a value of 121 rather
than 180 in equation (2).
No data are available for the adjustment of coefficients in turbulent flow. It is best to measure
them experimentally.
 Furthermore, the above theory does not account for:
(1) Channelling of the fluid
(2) Electrostatic forces between particles
(3) Agglomeration of particles
(4) Friction between the fluid and vessel walls
 Calculation of u
t
:
When the drag force exerted on a spherical particle by the upflowing gas > the gravity force
(based on apparent density) on the particle, the particle will be blown out of the bed.
i.e. blowout when:
g v F
g s p D
) ( =
But
p D t g D
A C u F = )
2
1
(
2
where:
C
D
= drag coefficient
A
p
= t d
p
2
/ 4 (projected area of particle)
g v C u
d
F
g s p D t g
p
D
) (
8
2
2
t
= =
2
1
3
) ( 4


.

\

=
g D
g s p
t
C
g d
u
(4)
For spherical particles, and Re < 0.4 (
p t g
d u
= Re )
Re
24
=
D
C
and eqn.(4) reduces to Stokes law:
Page 45 of 90
18
) (
2
p g s
t
gd
u
= (5)
For: 1 < Re < 10
3
, Trambouze et al (1984) give:
99 . 7 ln(Re)
43 . 69
50 . 5 ) ln(
+
+ =
D
C
For Re > 10
3
, C
D
= 0.43, and eqn.(4) reduces to Newtons law:
2
1
) ( 1 . 3


.

\

=
g
g s p
t
g d
u
(6)
 Fluidisation Regimes:
We can now consider fluidisation regimes for Geldart type A or B particles.
(see figures below)
(i) Key points for fluidisation regimes with coarse particles:
(1) Bubbles appear as soon as u
mf
is exceeded
(2) In turbulent regime, bubble life time is short (bubble burst). Bed looks quite uniform. (short
circuiting of gas through bubbles is less likely)
(3) u
t
and particle blowout coincide
(4) In fast fluidisation regime, there is net entrainment of solids
(5) In transport regime, there is solid flow in the direction of gas flow
(6) Carryover (entrainment) separates particles by size
Page 46 of 90
Page 47 of 90
(ii) Keypoints for fluidisation regimes with fine particles:
(1) Bubbles do not appear as soon as minimum fluidisation is reached; instead there is the
uniform expansion of the bed
(2) The bed is more coherent rather than with particles behaving independently
(3) Turbulent regime sets in well after u
0
exceeds u
t
of an individual particle. Thus we can
operate at higher u
0
(4) Carryover does not separate particles by size (i.e. a more cohesive bed)
 Heat and mass transfer in fluidised beds:
 Heat exchange can be through vessel walls or to an internal heat exchanger. For the latter, we
need to be more careful not to adversely effect the flow of fluidising gas, e.g. we can use
bayonet tubes. In some cases fluidising gas supplies or removes the reaction heat.
 Most heat transfer correlations assume a pseudosinglephase system, i.e. do not distinguish
between heat transfer from bubble and emulsion phases.
 For gas fluidisation, heat transfer is also dependent on the geometrical arrangement of the
vessel, and the type of fluidisation.
 Several correlations are available in the literature for heat transfer to wall or internal heat
exchanger. Usually they are in the form:
Nu = f(Re)
e.g. Dow & Jakob correlation for heat transfer to vessel walls:
8 . 0
25 . 0
17 . 0
65 . 0
) 1 (
55 . 0


.

\



.

\



.

\


.

\

= =
c
c
g t e
p g
p s
g
t
d u
C
C
dp
dt
L
dt d h
Nu
g
S
(i.e. Nu = f(Re
0.8
) )
 For heat transfer between gas and particles, Balakrishnan and Pei (1975) give a j
H
factor:
25 . 0
2
2
0
) 1 )( (
043 . 0


.

\

=
g
g s p
H
u
g d
j
c
But, in reality the above correlations for heat transfer is difficult to attain because heat transfer
rates between gas and particles is so large!
Kay & Nedderman (1979):
(Good news:)
Thus for all beds of depth greater than a few centimetres the gas in the particulate [emulsion]
phase may be assumed to be at the same temperature as the particles
(Bad news:)
By contrast, the gas in the bubble phase is not in contact with particles and need not be at the
same temperature. Consequently the mean temperature of the gas leaving the bed may differ
Page 48 of 90
from that of the particulates.
(Is the second statement strictly true?)
Page 49 of 90
4.2 Overview of Key Applications
As mentioned above, reactors involving fluidisation are useful for highly exothermic systems and/or
systems requiring close temperature control, e.g. oxidation reactions.
(When would close T control be needed, even if reaction is mildly exothermic or endothermic?)
In classical fluidised bed operation, the catalyst particles are retained in the bed, i.e. there is little
catalyst entrainment.
E.g.:
(1) Oxidation of naphthalene into phthalic anhydride
(2) Ammoxidation of propylene to acrylonitrile
(3) Oxychlorination of ethylene to ethylene dichloride
(4) Coal combustion (can inject limestone for the in situ capture of SO
2
)
(5) Roasting of ores
Even with classical fluidised beads, the region above the surface of bed still has some solid
concentration. This concentration becomes constant as we move further away from the surface, i.e.
We can apply cyclones above the TDH to recover and return the catalyst.
(Correlations are available to predict TDH, e.g. see Geldart).
 In transport reactors, the total amount of catalyst is entrained by gas
E.g.:
(1) FischerTropsch reactions (Sasol): production of hydrocarbons from synthesis gas (CO +
H
2
)
(2) Catalytic cracking, e.g. gasoline and diesel oil production from the cracking of crude oil (e.g.
BP oil, Mobil)
 Such reactions are associated with catalyst deactivation. The transport operation allows flow of
catalyst to a regeneration process (which may involve burning off the coke in a classical fluidised
bed!). The catalyst is then recirculated.
Constant solids
concentration Transport
disengagement height
(TDH)
Page 50 of 90
Furthermore, transport operation enables very short reaction spacetimes, which may be needed
to prevent overcracking, or a reduction in product selectivity.
 In fluid catalytic cracking (FCC) reactors (also referred to as cat crackers) reactions are
endothermic, but the combustion of coke in regeneration stage is used to reheat the particles, and
thus provide the required reaction heat.
(typically 12% (w/w) coke is reduced to 0.40.8% (w/w) in regeneration step)
Modern cat crackers use zeolite catalysts, which are highly active for cracking.
(see diagram below for FCC reactor)
 Usually there is some downflow of catalyst along the wall of the riser, which can upset the
desired product distribution.
The next generation of transport reactors may involve fluidised flow in the downward rather than
upward direction.
Page 51 of 90
4.3 Modelling of Fluidised Bed Reactors: NonTransport
4.3.1 TwoPhase Models
 Early models considered the fluidised bed as singlephase PFR, CSTR or ADPFR. This is
generally poor description of the process. (Why?)
 More accurate description is achieved through twophase model, with interchange between the
phases:
 Bubble phase ~ PFR
Emulsion phase ~ wellmixed or ADPFR
Twophase model requires 6 parameters:
(1) f
b
: fraction of bed occupied by bubbles (m
3
bubbles
/ m
3
bed
)
(2) f
e
: fraction of bed occupied by gas in the emulsion phase (m
3
emulsion gas
/ m
3
bed
)
(3) k
I
: gas interchange coefficient between bubble and emulsion phases (m
3
g
/ (m
3
bed
s))
(4) D
Ze
: dispersion coefficient in emulsion phase (m
2
/s)
(Dont need D
Ze
for wellmixed emulsion phase)
(5) g
b
: mass of solids in bubble phase (kg / m
3
bubbles
)
(6) g
e
: mass of solids in emulsion phase (kg / m
3
emulsion
)
Thus for isothermal fluidised bed with ADPFR in emulsion phase, the material balances give
(basis: mol / (m
3
bed
s) ):
Bubble phase:
0 ) (
'
= + +
Ab
e b
b
C on based
A b b A A I
A
b b
r g f C C k
dz
dC
u f (7)
(Analysis on units?)
Bubble
phase
u
0
, C
A0
u
b
, C
Ab
u
e
, C
Ae
Emulsion
phase
C
Ae

out
C
Ab

out
A
C
Page 52 of 90
Emulsion phase:
0 ) 1 ( ) (
'
2
2
= +
Ae
e b
e
e
e
C on based
A e b A A I
A
z e
A
e e
r g f C C k
dz
C d
D f
dz
dC
u f (8)
(Analysis on units?)
Also:
e b
A e e A b b A
C u f C u f C u + =
0
(9)
Boundary conditions for solution of above:
Bubble phase:
z = 0 C
Ab
= C
Ao
Emulsion phase:
0
) ( 0
0
= =
= =
dz
dC
L z
C C u
dz
dC
D z
e
e
e
e
A
A A e
A
z
 Simplification of the twophase model:
If u
b
>> u
e
, i.e. u
b
>> u
mf
, the emulsion phase will be ~ closed (relatively negligible inlet or outlet
flow)
Thus eqn.(8) reduces to:
'
) 1 ( ) (
A e b A A I
r g f C C k
e b
= (10)
(Also neglecting dispersion)
Eqn.(10) assumes a stagnant emulsion phase, but C
Ae
varies with the bed length (z).
E.g.:
1
st
order reaction with twophase model in which the emulsion phase ~ closed.
0 ) ( = + +
b e b
b
A b b A A I
A
b b
kC g f C C k
dz
dC
u f
and
0 ) 1 ( ) ( = =
e e b
A e b A A I
kC g f C C k
and
b e
A A
C C k =
where
k g f k
k
e b I
I
) 1 ( +
= k
(i.e. for k
I
>> k, C
Ae
~ C
Ab
;
for k
I
>> k, C
Ae
~ 0 )
0 ) ( = + +
b b b
b
A b b A A I
A
b b
kC g f C C k
dz
dC
u f k
0 ' = +
b
b
A
A
b b
C
dz
dC
u f k
where k g f k
b b I
+ = ) 1 ( ' k k
Page 53 of 90
Integrate with boundary conditions z = 0, C
Ab
= C
Ao
:
)
'
(
exp
0
b
b
b f
A
A
C
C
t k
= (11)
where
b
f
b
u
L
= t
Note: since there is gas outflow from the bubble phase only
A A
C C
b
=
 Estimation of parameters appearing in twophase model:
(i) u
b
:
r
b mf b
u u u u + = ) (
0
where u
br
is bubble rise velocity.
(Basis for above?)
Werther (1978) gives for a swarm of bubbles:
2
1
) ( g d u
b b
r
o =
where:
d
b
= bubble diameter
o = 0.64 for d
t
< 0.1m
= 1.6 d
t
0.4
for 0.1m < d
t
< 1.0m
= 1.6 for d
t
> 1.0m
Note: d
b
will depend on gas distributer design (see Werther for correlations). For Geldart A
particles, d
b
~ 10cm.
(ii) f
b
:
b
mf
b
u
u u
f
) (
0
= (Why?)
(
b
b
u
u
f
0
~ for u
b
>> u
mf
)
(iii) f
e
:
f b e
f f c = + (Why?)
(c
f
= voidage of fluidised bed)
(iv) L
f
and c
f
:
Since total volume of solids is constant:
bed packed
b mf mf f f
L L L ) 1 ( ) 1 ( ) 1 ( c c c = =
e.g.
b
f
mf
mf
f
f
L
L
= =
1
) 1 (
) 1 (
c
c
(Why?)
Page 54 of 90
(given f
b
& c
mf
(~0.4), L
f
& c
f
can be calculated)
(v) D
ze
:
Correlations available in literature, e.g.:
) , (
0 b z
d u f D
e
=
(vi) u
e
:
mf
mf
e
u
u
c
= (Why?)
(vii) g
b
and g
e
:
volume bed fluidised
mass solid total
g f g f
e b b b
= + ) 1 (
i.e.
f
e b b b
L A
M
g f g f
= + ) 1 (
Note: typically g
b
~ 0
(viii) k
I
:
For twophase models, k
I
is often used as a fitting parameter such that the model agrees with
plant data.
Page 55 of 90
4.3.2 ThreePhase (Hydrodynamic) Models
 Davidson & Harrison (1963): they worked on gas flow in the vicinity of a rising bubble in a
fluidised bed of fine particles:
(in large particles: cloudless bubble)
 Thus we can have gas interchange from bubble to cloud, then from cloud to emulsion. i.e.
sequential steps.
i.e.
 Again, different mixing regimes in phases can be assumed
 Kunii and Levenspiel model (KL): assumes emulsion phase with no net gas flow (closed).
(Usually achieved for 6
0
>
mf
u
u
)
e.g. KL model for 1
st
order reaction.
Material balances:
(i) bubble phase:
0 ) ( = + +
b C b b
b
A b b A A I
A
b b
kC g f C C k
dz
dC
u f k (12)
(ii) emulsion phase:
e C e
A e c b Ae A I
kC g f f C C k ) 1 ( ) ( = k (13)
i.e.
g
e
kC
Ae
: (mol / (m
3
emulsion
s) )
(1f
b
f
c
): (m
3
emulsion
/ m
3
bed
)
bubble cloud emulsion
C
Ab
C
Ac
C
Ae
k
Ib
k
Ie
Page 56 of 90
(iii) cloud phase (link bubble & emulsion phases):
C e C b C b b
A c c A A I A A I
kC g f C C k C C k
'
) ( ) ( + = k k (14)
In eqn.(14):
f
c
: m
3
cloud
/ m
3
bed
g
c
: kg / m
3
cloud
) 1 (
~
fraction bubble
b
density bulk
b
e
f
g
=
Note: from experimental correlation, Partridge and Rowe give:
) ( 17 . 0
17 . 1
'
e
b
b c
u
u
f f
+
=
 Using eqn.s (13) & (14), we can express C
Ac
in eqn.(12) in terms of C
Ab
(c.f. previous example).
This gives:
b
b
A
A
b
C
dz
dC
u k = (15)
where (believe it or not):














.

\

+
+
+
+ =
b
c b e I
b
b
c c I
b
b
f
f f g k
kf
f
f g k
kf
g k
e
b
) 1 (
1
1
1
1
'
'
k
(k effective rate constant for a threephase fluidised bed model, i.e. KL rate constant)
Integration of (15) with boundary conditions z = 0, C
Ab
= C
Ao
:
) (
exp
0 0
b
b
A
A
A
A
C
C
C
C
kt
= =
(16)
where
b
f
b
u
L
= t
 Note: the work of past investigators (e.g. KL, Partridge & Rowe, Davidson & Harrison) has led
to equations for providing predictions for k
Ib
& k
Ie
based on hydrodynamics.
(see Levenspiel (1999))
 3phase models are found to give better prediction of reactor performance.
Page 57 of 90
4.4 Modelling of Transport (Riser) Reactors
 E.g. FCC processes: fast reactions (small t needed) and rapid catalyst deactivation.
 Velocity of solids ~ velocity of gas, i.e. no slip velocity.
 Usually employ fine solids, Thus q ~ 1
 For no catalyst deactivation, riser is very much like a pseudohomogeneous PFR, but
bed packed
b
c c > :
 Calculation of c:
p
s
b
g
m
Au
Au
m
m
c

+
=
0
0
3
3
) (
where:
p
= pellet density (kg / m
3
pellet
)
p
s
u
A m
c
0
) / (
1
1

+
=
(17)
i.e. when
S
S m /

<< u
0
, c 1
when
S
S m /

>> u
0
, c 0
 Thus for no catalyst deactivation:
p A
A
r
dz
dC
u c) 1 (
'
0
= (18)
Unit analysis on R.H.S. of (18):
s m
mol
m
kg
m
m
s kg
mol
b p b
p
3 3 3
3
. . . .
.
. . . .
.
. . . .
.
. . . .
.
. . . .
.
. . . .
.
. . . .
.
. . . .
.
. . . .
.
solid S m

(kg/s) gas u
0
A
c
Page 58 of 90
 Catalyst deactivation in FCC is believed to arise from coke deposition and the adsorption of
certain (basic) species present in the feed.
Will result in reduction in the reaction rate(s); reaction rates will decrease with time. Can be
described using deactivation function.
Deactivation function, u:
) (
] 0 [
] [
'
'
t f
r
t r
A
A
A
= = u (19)
e.g. (1) u = 1  ot
(2) u = exp
( ot)
Then, eqn.(18) becomes:
p A A
A
r
dz
dC
u c) 1 (
'
0
u = (20)
t in eqn.(19) represents the amount of time the catalyst has spent in the riser,
i.e.
0
u
z
t =
(for no slip )
 Sometimes u is given as a function of the coke concentration on the catalyst pellets. It is
practical to express this concentration in terms of:
( kg
coke
/ kg
catalyst
)
C
C
The rate of formation of coke is then:
C
r kg
coke
/ (kg
cat
s)
C
r can itself be deactivated as coke is produced!
Thus, material balance for coke deposition:
) 1 (
c u =

p C c
C
S
r
dz
C d
A
m
(22)
(Units analysis on above?)
(Eqn.(22) has to be solved simultaneously with eqn.(20) for this case)
 Finally, the energy balance for an adiabatic riser can be written as:
) 1 ( )] ( ) ( [
'
c A u + A u =
+
 
p C C c A A A
p
s
p
g
H r H r
dz
dT
A
C m C m
S g
(23)
where C
pg
& C
ps
are the specific heat capacity of the gas and solid respectively (kJ/(kg K)) and
g m

the mass flow rate of gas (kg/s).
(
g
S M
RT
P u A
m =

0
0 0
)
What are the additional assumptions of the riser model given by eqn.s (20), (22) and (23)?
Page 59 of 90
5 Multiphase Reactors
5.1 Background
 Two or more phases are needed to carry out the reaction (i.e. gasliquid; liquidliquid).
 Majority of multiphase reactors involve gas and liquid phases in contact with a solid.
Solid may be:
 Catalyst particles dispersed in the liquid phase, e.g. slurry reactor
 Packing for liquid distribution, e.g. packed bed absorber (CO
2
in MEA)
 Packing for liquid distribution and catalyst support, e.g. trickle bed reactor, packed bubble
reactor
 Plates for liquidgas contact (c.f. distillation column)
 Reactors can also be classified in terms of which phase is continuous and which is dispersed.
e.g.:
Liquid phase continuous; gas phase dispersed:
Bubble reactor, slurry reactor, fermentation vessel.
Liquid phase dispersed; gas phase continuous:
tower (also trickle bed and packed bed reactors).
G
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
G
L
Page 60 of 90
Liquid phase continuous; gas phase continuous:
Falling film (wetted wall) reactor (When would such operation be useful?)
 If the main mass transfer resistance is located in liquid: use dispersed gas phase and continuous
liquid phase.
 If the main mass transfer resistance is located in gas: use continuous gas phase and dispersed
liquid phase.
 Residence time of reactant and heat transfer considerations will also dictate the type of reactor,
e.g.:
 Plate columns can achieve long contact time between the gas and liquid, but poor temperature
control.
 Stirred tanks (e.g. bubble or slurry reactors) will have large L:G ratio, but yet cope with high G
flowrates and have good temperature control.
 Reactors can have co or counter current flow of G and L to utilise driving forces for mass and
heat transfer.
 Where reactors are principally employed for gas purification, they are referred to as absorbers.
 The theories for GL and LL systems are similar: the former uses Henrys law constant to
describe equilibrium distribution of a component in the G & L phases, the latter uses a partition
coefficient. Our analysis will consider GL systems.
 For the mathematical description of multiphase reactors, two situations arise depending on the
relative rates of reaction and mass transfer:
(i) reaction rate >> mass transfer rate:
E.g. fluidfluid reactions involving a homogeneous (soluble) reaction, or requiring no catalyst. In
this case, mass transfer across, say, the liquid film (see Whitmans twofilm theory) will be
accelerated due to the reaction.
(ii) reaction rate << mass transfer rate:
E.g. Fluidfluid reactions involving a catalyst (suspended or fixed). In this case, mass transfer
across the liquid film will not be accelerated due to the reaction.
Page 61 of 90
5.2 Review of TwoFilm Theory
 Twofilm theory assumes that stagnant layers (films) exist in both the G and L phases along the
interface.
All resistance to mass transfer are then assumed to be located in the G and L films.
No resistance at interface, such that Henrys law is satisfied:
i i
A A
HC P = (1)
Thus:
 We consider the reactor scheme:
aA(gl) + bB(l) Products (l)
i.e. A dissolved in liquid and reacts with nonvolatile B to produce nonvolatile product(s).
 When reaction rate is relatively small comparing to mass transfer rate (usual assumption for
solidcatalysed reactions):
A
y
A A
N N N
L
= =
0
and
) ( ) (
i L i i
A A L A A G A
C C k P P k N = = (2)
i.e. negligible reaction in liquid film, and no mass transfer enhancement due to reaction.
 When reaction rate is relatively large comparing to mass transfer rate (usual assumption for fluid
fluid reactions which require no catalyst, or use a homogeneous catalyst):
L
y
A A
N N =
0
and
) (
0 i L i
A A L A
C C k N >
i.e. reaction is in liquid film, and there is mass transfer enhancement due to reaction.
 Calculation of N
A

o
(N
A

yL
) (see Separation course):
Solution of diffusion equation in liquid film, i.e.
N
A

o
G L
(reaction in this
phase only)
P
A
P
Ai
C
Ai
C
AL
N
A
= N
A

o
N
A

yL
0
y
L
.
.
. .
Page 62 of 90
A
A
A
r
dy
C d
D =
2
2
Boundary conditions: y = 0, C
A
= C
Ai
y = y
L
, C
A
= C
AL
where: D
A
= molecular diffusivity of A in liquid film (m
2
L
/s)
Then:
y
A
A
y
A
dy
dC
D N =
 E.g.:
1
st
order irreversible reaction:
A A
kC r = (mol / (m
3
L
s))
or
A B A
C kC r ) ( = with C
B
in excess
Then:
sinh
) ( sinh ]) 1 [ ( sinh
L
A
L
A
A
y
y
C
y
y
C
C
L i
+
=
where:
= Hatta number
2
1


.

\

=
A
L
D
k
y (3)
(c.f. Thiele modulus!)
Since
L
A
L
y
D
k :
2
1


.

\

=
L
A
k
kD
(4)
Thus, for this case:
) cosh (
sinh
0 L i
A A
L
A
C C
k
N =
(5a)
and
) cosh (
sinh
L i
L
A A
L
y
A
C C
k
N = (5b)
Note: when 0 (k very small):
) (
0 L i
L
A A L
y
A A
C C k N N = =
(Why arent moles conserved such that
L
y
A A
N N =
0
for all ?)
 Enhancement Factor:
Sometimes it is convenient to work in terms of an enhancement factor to calculate N
A
:
Page 63 of 90
) (
0
L i
A A L
y
A
C C k
N
E
=
(6)
i.e. E = (flux with reaction enhancement)/(flux with no reaction enhancement)
Thus, if E is known, we can model
0
A
N as:
) (
0 L i
A A L A
C C Ek N =
rather than using a complex equation.
 Simple expressions can be derived for E for certain reactions:
E.g.:
(i) 1
st
order irreversible reaction:
Substitute eqn.(5a) into (6):
)
cosh
1
1 (
tanh
=
i
b
A
A
C
C
E
For C
Ab
<< C
Ai
(often assumed):
tanh
= E (7)
(c.f.


q tanh
1
= for slab)
(Why is this similar result not surprising?)
Then: ) (
0 i
A L A
C Ek N = where
L
A
C is not involved
(ii) Instantaneous and irreversible reaction between A and B:
Can show for this case:
) 1 (
0
i
L
i
A A
B B
A L A
C bD
C aD
C k N + = (8)
Substitution of (8) into (6) and noting that C
AL
= 0:
G L
P
A
P
Ai
C
Ai
C
BL
reaction plane
.
.
C
A
=C
B
=0
Page 64 of 90
i
L
A A
B B
ins
C bD
C aD
E + =1 (9)
i.e. E
ins
f()
 Plots of eqn.s (7) & (9) are given by van Krevelen and Hoftijzer (see figure below).
 Criteria for speed of reaction, and approximations for E:
(1) < 0.2:
slow reaction (mass transfer non limiting)
E ~ 1
(2) 0.2 < < 2:
intermediate reaction
E ~ 1 +
2
/3 (for 1
st
& 2
nd
order reactions)
(3) > 2:
fast reaction
(i) > 5E
ins
: instantaneous reaction
(ii) < 0.2E
ins
: E ~ (for 1
st
& 2
nd
order reactions)
Van Krevelen and Hoftijzer plot:
1. Lines below dashed line: instantaneous reaction.
2. E ~ for > 2 and < 0.2E
ins
E~1 (for <0.2)
E
ins
1
E = /(tanh) ~
Page 65 of 90
3. Curves between dashed line and E = /(tanh): from numerical simulation.
4. E ~ 1 +
2
/3 for 0.2 < < 2
 The above expressions for N
A

o
do not account for gasphase mass transfer resistance. Since
transport processes are in series, i.e.
Then for C
AL
~ 0:
) ( ) (
i i
A L A A G
C Ek P P k = where
i
A
C is not involved
i i
A L A A G
C Ek HC P k = ) (
H
k
Ek
P
C
G
L
A
A
i
+
=
H
k
Ek
P Ek
C Ek N
G
L
A L
A L A
i
+
= =
0
)
`
+
=
L G
A
A
Ek
H
k
P
N
1
0
(10)
G
P
A
P
Ai
C
Ai
gas film
. N
A

o
L
resistance in series
Page 66 of 90
5.3 General Design Models for Multiphase Reactors
 As mentioned above, for the case of fluidfluid reaction with no catalyst, we need to account for
L
u
A A
N N =
0
.
 Design models need to account for plug flow or wellmixed flow of the different phases.
 Although there will be specific design issues depending on the actual type of reactor (e.g.
flooding, entrainment, weeping, bubble size control, etc ), the underlying reaction,
absorption and flow phenomena can be described in a relatively simple way.
E.g.: The simplest model of a stirred vessel involving multiple phases: all phases completely
mixed. Thus just like a CSTR.
(What models could be used for: (i) a bubble column, (ii) a spray tower, (iii) a trickle bed
reactor?)
5.3.1 Gas & Liquid Phases Completely Mixed
 E.g. Slurry reactor, liquidliquid reactor in a stirred vessel
(Basis: mol/s, single reaction: key reactant A)
 Gas phase material balance:
) 1 ( ) (
0
g v A A A
V a N y y G
out in
c = (11)
a
v
: (m
2
gasliq interface)/(m
3
liquid)
c
g
: (m
3
gas)/( m
3
gas+liquid) (1  c
g
?)
i.e. R.H.S. of eqn.(11):
s
mol
m
m
m
m
m
s m
mol
G L
L
G L
L
i
i
=
+
+
3
3
3
3
2
2
 Liquid phase material balance:
Since reaction in liquid film is already accounted for by N
A

yL
, we need to consider reaction in
the remaining liquid volume, i.e.:
V
L
= liquid volume neglecting liquid film
G (mol/s)
y
Ain
G
y
Aout
L (mol/s)
x
Ain
L
x
Aout
V (m
3
)
Page 67 of 90
) {
) 1 ( ) ( ) 1 (
3
'
3
,
3
'
3
3
3
,
3
3
3
3
L film L L
m
m
m
g L v
m
m
m
g L
m m m
V y a V V
L
L
film L
L
m
G L
L
G L
=
=
+
+
_ _
_
c c
) 1 ( ) 1 (
' L v g L
y a V V = c (12)
(y
L
can be attained from:
L
A
L
k
D
y = )
Thus, liquid phase material balance can be written as:
) 1 ( ) (
' g v
y
A A A A L
V a N x x L r V
L
in out
c = + (13)
(r
A
in mol/(m
3
s).)
 Overall material balance:
E.g. Given aA + bB P then:
mol A consumed =
b
a
mol B consumed
Thus:
) ( ) (
out in out in
B B A A
x x L
b
a
y y G = (14)
 Degrees of freedom analysis on well mixed model (eqn.s (11)(14)):
E.g.:
Given: y
Ain
, x
Ain
, x
Bin
, G, L, V, a
v
, c
g
Unknowns = 6: y
Aout
, x
Aout
, x
Bout
, V
L
, N
A

o
, N
A

yL
Equations = 4: i.e. eqn.s (11)(14)
Thus require 2 more equations, i.e. for N
A

o
& N
A

yL
; e.g. eqn.s (5a) & (5b).
But N
A
= f(C
Ai
, C
AL
)
where
out
out
L
A L
L
A
A
x
L
Lx
C = =
) / (
(
L
: liquid density, mol/m
3
)
H
P
C
i
i
A
A
=
If gasfilm resistance is negligible,
out i
A A A
y P P P = =
Otherwise use:
0
) (
A A A G
N P P k
i
=
Note: In a gas purification problem, y
Aout
& y
Ain
may be specified, in which case we can solve for
L or V.
 For slow reaction (e.g. with a solid catalyst):
) (
0 out in
L
A A L L
y
A A
x x k N N = =
Page 68 of 90
 For very rapid reaction: reaction completed in liquid film x
Aout
= 0.
0 =
L
y
A
N
and we dont need eqn.(13)
Can use E to calculate
0
A
N for this case.
Example: Wellmixed GL Reaction:
Liquid phase oxylene oxidation to omethylbenzoic acid by means of air in a CSTR.
L
= 7.1 kmol/m
3
D
A
(O
2
through xylene) = 5.2 x 10
6
m
2
/hr
H
A
= 126.6 m
3
bar/kmol
a
v
(1 c
g
) = 2089 m
2
i
/m
3
L+G
(
'
v
a )
c
g
= 0.336 m
3
G
/m
3
L+G
k
L
= 1.485 m
3
L
/(m
2
i
hr)
Calculate V, x
Aout
, and y
Aout
for a desired conversion of oxylene of 16%.
Solution:
If 16% oxylene conversion:
in
out in
B
B B
Lx
x x L ) (
16 . 0
=
hr kmol x x L
out in
B B
/ 5 . 27 ) ( = and x
Bout
= 0.84
Also: m
k
D
y
L
A
L
6
6
10 5 . 3
485 . 1
10 2 . 5
= =
r
A
= 3.6 x 10
3
C
A
= 3.6 x 10
3
(
L
x
Aout
)
Material balances:
(1) V a N y y G
g v A A A
out in
) 1 ( ) (
0
c =
(2) V a N Lx r y a V
g v
y
A A A L v g
L
out
) 1 ( ) 1 )( 1 ( c c = +
(3) ) ( ) (
out in out out in
B B A A A
x x L
b
a
Lx y y G =
Thus, given N
A

o
& N
A

yL
: 3 unknowns (V, x
Aout
& y
Aout
) & 3 equations
Page 69 of 90
For a 1
st
order irreversible reaction, equations (5a) & (5b) are applicable:
) cosh (
sinh
0 L i
A A
L
A
C C
k
N =
) cosh (
sinh
L i
L
A A
L
y
A
C C
k
N =
where
L
A
k
kD
2
1
) (
=
i.e. ) 2 . 0 ( 10 2 . 9
485 . 1
) 10 2 . 5 10 6 . 3 (
2
2
1
6 3
< =
=
Therefore sinh ~ , cosh ~ 1
A A A L
y
A A
N C C k N N
L i
L
= = = ) (
0
Neglecting gas phase mass transfer resistance:
A
A
A
A
A
H
y P
H
P
C
out out
i
= =
) (
out
out
A L
A
A
L A
x
H
y P
k N
=
Thus, neglect a
v
& y
L
term!
Can now solve 3 material balance equations to give:
V = 6.78 m
3
x
Aout
= 3.6 x 10
4
y
Aout
= 4.15 x 10
2
(q.e.d.)
Note:
Stirrer design and speed will influence bubble diameter, d
b
.
d
b
in tern will influence a
v
, i.e.
b
g
g v
d
a
c
c
6
) 1 ( = (Why?)
k
L
can be obtained from correlation.
Page 70 of 90
5.3.2 Gas & Liquid Phases in Plug Flow
 E.g. packed tower
Consider a differential volume in the tower:
Also: use a
v
(m
2
i
) rather than a
v
(m
2
i
/m
3
L
) since correlations are available for a
v
for packed
towers.
 Gas phase material balance:
'
0
v A
A
a N
dV
dy
G = (15)
 Liquid phase material balance:
A L v g v
y
A
A
r y a a N
dV
dx
L
L
) 1 (
' '
= c (16)
 Overall material balance:
Can carry out balance either at top of column or at bottom, depending on what inlet/outlet
conditions are specified, i.e.
top: ) ( ) ( ) (
B B A A A A
x x L
b
a
x x L y y G
in in out
= (17a)
or bottom: ) ( ) ( ) (
out out in
B B A A A A
x x L
b
a
x x L y y G = + (17b)
How would the above model, i.e. eqn.s (15)(17) be modified for cocurrent operation?
How would the above model boundary conditions be simplified for complete reaction in the film?
Why is the choice of packing so important?
What are some of the hydrodynamic issues which we need to be careful about?
N
A
dV
G, y
A
L, x
A
x
Ain
y
Aout
y
Ain
x
Aout
V
Page 71 of 90
5.3.3 Gas Phase in Plug Flow, Liquid Phase Completely Mixed
 E.g. bubble column; slurry reactor (but with bubble flowing straight through)
(c.f. fluidized bed reactor)
 Gas phase material balance:
Eqn.(15) is again applicable:
'
0
v A
A
a N
dV
dy
G = (15)
 Liquid phase material balance:
Since bubble or gas p hase is in plug flow, N
A

o
& N
A

yL
will vary with height. Thus the
calculation of the total moles of A transferred from gas to liquid requires an integration.
Therefore:
) (
'
0
'
0 in out
A A A L
V
v A
x x L r V dV a N + =
}
(18)
(c.f. eqn.(13))
 Overall material balance:
Eqn.(14) is again applicable:
) ( ) ( ) (
out in out in out in
B B A A A A
x x L
b
a
x x L y y G = + (14)
5.3.4 Effective Diffusion Model
Similar to plug flow model (of gas, liquid, or gas & liquid phases) but a dispersion term is used to
account for some intermediate degree of mixing.
(see Froment & Bischoff, p. 608)
Page 72 of 90
5.4 Simplifications to Multiphase Design Models
5.4.1 Instantaneous Reactions
C
AL
= 0 (x
A
= 0)
N
A

yL
= 0
N
A

o
= E
ins
k
L
C
Ai
E.g.: gas & liquid phases wellmixed; neglect gasside mass transfer resistances:
Eqn.(11):
) 1 ( ) (
g v A L ins A A
V a C k E y y G
i out in
c =
(
H
y P
H
P
C
out i
i
A A
A
= = )
GH
V a P k E
y
y
g v L ins
A
A
out
in
) 1 (
1
c
+ =
Eqn.(13): not needed
Eqn.(14): ) ( ) (
out in out in
B B A A
x x L
b
a
y y G =
5.4.2 Very Fast Reactions
For reaction that is essentially completed in liquid film:
C
AL
~ 0
N
A

yL
= 0
Thus, equations in section 5.4.1 are again applicable, but need to use E rather than E
ins
.
5.4.3 Slow Reactions
) ( ) (
0 i L i
L
A A G A A L
y
A A A
P P k C C k N N N = = = =
Noting that P
Ai
= HC
Ai
and eliminating C
Ai
using k
L
& k
G
terms:
L G
A A
A
k
H
k
HC P
N
L
+
=
1
(19)
(c.f. eqn.(10))
E.g. Gas and liquid phases wellmixed
Eqn.s (11), (13), & (14) as before but
) 1 (
' g L L
V V V c = =
A
y
A A
N N N
L
= =
0
Page 73 of 90
Thus:
Eqn.(11):
) 1 ( ) (
g v A A A
V a N y y G
out in
c =
Eqn.(13):
) 1 ( ) ( ) 1 (
g v A A A A g
V a N x x L r V
out
c c = +
Eqn.(14): unchanged
5.4.4 Solid Catalyzed Reactions
For relatively slow reaction,
L
y
A A
N N =
0
We may need to account for mass transfer resistances associated with catalyst pellets or catalytic
surface.
E.g.
 r
A
q
G
r
A
 r
A
q r
A
(if externalfilm resistances are negligible)
Exercise:
Given q and k
mc
, write down the appropriate form of eqn.(13). What additional equation is needed?
5.4.5 Resistances in Series Approximation: GasLiquidSolid Reactions
(see also Levenspiel (1999))
 Consider reaction:
aA
(g)
+ bB
(l)
P
(l)
but 1
st
order with respect to A & B:
B A A
C kC r =
'
(mol/(kg s))
(also:
' '
A B
r
a
b
r = )
 Assume reaction and transport steps are in series:
R
A
: reaction rate (mol/(m
3
reactor
s))
G
(A)
(ii)
L
(i)
(B)
cat.
(iii)
(iv)
.
Page 74 of 90
i.e.
R
A
) 1 (
'
g L A
g r c q = (iv)
) (
S
A A C m
S L C
C C a k = (iii)
) (
'
L i
A A v L
C C a k = (ii)
) (
'
i
A A v G
P P a k = (i)
where:
S
B
S
A A
S S
C kC r =
'
g
L
: kg
cat
/m
3
L
a
c
: m
2
cat surface
/m
3
reactor
k
mc
: m
L
/s
Why is the above description of R
A
wrong?
When would it be ~ true?
Noting that P
Ai
= HC
Ai
in (i), and eliminating C
Ai,
C
AL
,
S
A
S
C using (i)(iv), i.e.
H a k
a k
i iii
a k
a k
ii
v G
C m
v L
C m
C C
' '
) ( ) ( ) ( +


.

\

+
Gives:
S
A
A
v G v L C m
A
S
C
C
H
P
H a k a k a k
R =


.

\

+ +
' '
1 1 1
Then substitute for
S
A
S
C using (iv):
A A A
P k R = (20)
where:
) 1 ( ) (
1 1 1 1
' '
g L
S
B v G v L C m
A
g kC
H
a k a k a k
k
S C
c q
+ + + = (21)
i.e. resistances in series (quantitatively useful)
If we also assume:
C
BL
>> C
AL
i.e. Pure liquid B and slightly soluble A, then:
L S
B
S
B
C C = = constant
Thus
A
k = constant, since ) (
S
B
S
kC in eqn.(21) is constant
 Design applications of
A
k is straightforward:
(i) Gas and liquid phase are wellmixed:
) (
B Bo A A Ao
X n
b
a
V r X n = =
(X
A
& X
B
denote conversions!)
Page 75 of 90
i.e. V X P k V P k X n
A Ao A A A A Ao
) 1 ( = =
t ) (
1
RT k
X
X
A
A
A
=
where
g
V
V
0
= t
(ii) Gas in plug flow (and any flow of liquid phase (Why?))
A
A
Ao
R
dV
dX
n =
t ) (
exp 1
RT k
A
A
X
=
5.4.6 Resistances in Series Approximation: GasLiquid Reactions
Similar method to above, but no solid resistances and need to account for absorption enhancement.
R
A
3
3
3
) 1 (
reactor
L
L
L L
m
m
g
s m
mol
B A v
C C k c =
) (
'
L i
A A v L
C C a Ek =
) (
'
i
A A v G
P P a k =
(Again, why is this wrong? When would it be ok?)
Can then show:
A A A
P k R = (22)
where for this case:
) 1 )( (
1 1
' '
g B v v L v G
A
L
C k
H
a Ek
H
a k
k
c
+ + =
 For C
BL
in excess, i.e. pure liquid B in feed, R
A
again is constant if E is constant (or assumed ~
constant)
 Note for very fast or instantaneous reaction (complete reaction in liquid film):
v
k , and 0 ~
L
A
C
' '
1 1
v L v G
A
a Ek
H
a k
k
+ =
and transport steps are in series.
We could have, or course, derived this model from section 5.4.1 (or 5.4.2) but using:
) (
0
A A G A L ins A
P P k C k E N
i i
= =
Page 76 of 90
Example: Wellmixed GL reaction:
Reconsider oxylene oxidation to omethylbenzoic acid by means of air in a CSTR, but this time
using eqn.(21).
=
A A
B B
P k
b
a
X n
V m
3
Calculations consistent for this case because:
(i) C
BL
>> C
AL
(pure liquid B as feed)
and
(ii) C
AL
(i.e. x
Aout
) ~ 0, thus L(x
Ain
x
Aout
) ~ 0
Note: y
Aout
is specified here (calculated using previous (nonapproximate) method). If this was not
specified then:
) 1 (
0
A A A
X P P =
But:
B B A A
X n
b
a
X n
0 0
=
8023 . 0 16 . 0
21 . 0 245
172
5 . 1
0
0
=
= =
B
A
B
A
X
n
n
b
a
X
1 . 57284 ) 8023 . 0 1 )( 21 . 0 ( 8 . 13 = =
A
P Pa
Page 77 of 90
5.5 Factors in Selecting a GasLiquid Contactor
 Mass transfer driving forces in towers are higher than in agitators, but for high L/G rations
agitators are better.
Usually for a packed tower: L/G ~ 10 at 1bar.
 Liquid droplet vs. gas bubble:
Droplet:
 k
G
high (the gas flow around droplet is high)
 k
L
low (liquid is stagnant in droplet)
Bubble:
 k
G
low (the gas is stagnant in bubble)
 k
L
high (there is relative motion between liquid film and bulk liquid)
i.e.
(1) Dont use spray tower if k
L
is low
(2) Dont use bubble column if k
G
is low
(3) For very soluble gases (H small), there is gasfilm mass transfer controls, i.e. k
G
low. Thus
better use a dispersed liquid (droplet) or packet towers. Should avoid bubble contactors.
(4) For gas of low solubility (H high), there is liquidfilm mass transfer controls, i.e. k
L
low.
Should avoid spray contactors.
(see also Levenspiel (1999), for typical characteristics of GL contactors in terms of:
'
v
a , (1c
g
),
capacity)
Page 78 of 90
6 NonCatalytic FluidSolid Reactions
E.g.
 coal gasification/burning
 ore processing
 iron production (blast furnace)
 regeneration of coked catalysts
 activation of catalysts (i.e. reduction or oxidation)
 Si oxidation to SiO
2
for fabrication of microelectronic devices
 pharmacokinetic processes
 Can classify different reactions into two general types:
(i) Total particle dissolution
i.e. particle is being completely consumed and thus shrinking with time.
E.g. tablet dissolution, coal gasification
(ii) Shrinking core,
i.e. overall particle size remains unchanged, but the reactive components within the particle is
decreasing in concentrations. Thus a shrinking core of reactive material.
E.g. catalyst regeneration activation
Regeneration of coked catalyst pellet
1.0
F
r
a
c
t
i
o
n
o
f
c
o
k
e
b
u
r
n
e
d
time
0
Page 79 of 90
6.1 Total Particle Dissolution
 Consider A diffusing to surface to react with solid B:
At the surface: aA
(g/l)
+ bB
(s)
P
(g/l)
Reactions of this type are usually zeroorder in B, and firstorder in A:
S
A S
S
A
C k r = (mol/(m
2
s))
At any time:
) (
S
A A m
S
A S A
C C k C k N
C
= =
A
m S
m
S
A
C
k k
k
C
C
C
+
=
Thus:
A
S
A A
C k r N = =
0
(1)
where:
S m
k k k
C
1 1 1
0
= = (2)
(c.f. eqn.s (4) and (5) in section 2)
 Effect of d
p
on k
mc
:
For flow around a spherical pellet:
3
1
2
1
Re 6 . 0 2 Sc Sh + = (Frssling correlation)


.

\

= = =
A g
p g
A
p m
D
Sc
ud
D
d k
Sh
C
, Re ,
i.e. k
mc
= f(d
p
) = f(t)
For small particles and/or low u:
Sh ~ 2, i.e. k
mc
= 2D
A
/d
p
Thus, eqn.(2) becomes:
S p A
k d D k
1
/ 2
1 1
0
= = (3)
i.e. equal resistances when
*
2
p
S
A
p
d
k
D
d = (4)
when
*
p p
d d >> : mass transfer controlled
B N
A
(mol/(m
2
s))
S
A
r
.
Page 80 of 90
when
*
p p
d d << : reaction rate controlled
 Now consider solid material balance:
accumulation = consumption (basis: mol/s)
p
S
B B p
A r V
dt
d
= ) (
(
3
2
3
; ;
6
p
B
B p p
p
p
m
mol
d A
d
V = = = t
t
)
For a = b = 1,
S
A
S
B
r r = :
B
A
B
S
A
p
C k r
d
dt
d
0
2 2
) (
=
= (5)
Given excess C
A
(C
A
= constant), and eqn.(3) for k
0
, eqn.(5) can be integrated with boundary
conditions:
d
p
= d
po
at t = t
o
to give:
B
A S
p
p p p p
t C k
d
d d d d
=


.

\

+
*
2
) (
1
2
0 0
(6)
Therefore, for complete particle dissolution (d
p
= 0):


.

\

+ =
*
2
1
2
0
0
p
p
p
A S
B
d
d
d
C k
t
(7)
Note: For agitated systems and/or systems involving complex kinetics, numerical integration of
eqn.(5) may be necessary.
Page 81 of 90
6.2 Shrinking Core Model
 Demonstration of shrinking core model through application to catalyst regeneration (i.e.
decoking):
i.e. O
2
diffusion through externalfilm and shell to core surface, where a rapid oxidation of the
carbonaceous material occurs.
Diffusion through shell is usually ratecontrolling. (Why?)
 Although core is shrinking with time, at any instant we can assume the O
2
concentration profile
is the shell to be a steady state profile, i.e. quasisteady state (QSS) assumption.
(Why do we expect QSS to be ~ true?)
(Bischoff (1963, 1965) shows that QSS is true for:
3
10
<
B
A
C
)
 Consider diffusion equation for O
2
transport through (spherical) shell:
(i.e. eqn.(29) in section 2)
0 4 ) (
1
2 2
2
= = r r
dr
dC
D r
dr
d
r
S
A
A
e
A
t
For constant D
A
:
0 ) (
2
=
dr
dC
r
dr
d
A
(8)
Boundary conditions for eqn.(8):
r = R
o
, C
A
= C
Ao
r = R, C
A
= 0
(What are the assumptions here?)
Note: R = f(t) but will assume constant in establishing C
A
profile in the shell (QSS assumption).
Integration of eqn.(8) twice, using the above boundary conditions gives:
0
1 1
1 1
0
R R
r R
C
C
A
A
=
(9)
B
N
A
(mol/(m
2
s))
.
shell
core
A
(e.g. O
2
)
R R
o r
Page 82 of 90
i.e.:
Also:
S
A
R r
A
e
R r
A
r
dr
dC
D N
A
= =
=
=
S
A
A e
r
R
R R
C D
A
=


.

\

2
0
1 1
0
(10)
 Now consider solid (carbon) material balance:
accumulation = consumption (basis: mol/s)
2
2
0
4
1 1
) (
0
R
R
R R
C D
f V
dt
d A e
B B core
A
t


.

\

=
where:
3
3
4
r V
core
t =
=
B
molarity of coke (
3
coke
m
mol
)
f
B
=
3
3
pellet
coke
m
m


.

\

=
R R
R
R f
C D
dt
dR
B B
A e
A
0
0
0
0
(11)
Boundary conditions for eqn.(11):
R = R
o
at t = 0
Integration gives:
1.0
R
0
A
A
C
C
r R
0
0.0
core shell bulk
Page 83 of 90


.

\



.

\

+


.

\

=
3
0
2
0
2
0
2 3 1
6
0
R
R
R
R
C D
R f
t
A e
B B
A
(12)
Thus, time required for complete decoking (R = 0):
0
6
2
0
A e
B B
C D
R f
t
A
= (13)
E.g. Decoking of fluidised catalytic cracking (FCC) catalyst:
~ 2% (w/w) coke on 2mm diameter pellets:
(
B
f
B
): mol
coke
/m
3
p
3 3
2300
012 . 0
1
1400 02 . 0
p coke
coke
p
p
p
coke
B B
m
mol
kg
mol
m
kg
kg
kg
f = =
Regeneration with 5% O
2
at ~ 1000K and 1bar; D
eA
~ 3 x 10
5
m
2
/s:
) 6 . 0 )( 10 3 ( 6
) 10 1 ( 2300
5
2 3
= t
s t 20 ~
Page 84 of 90
6.3 Reactor Design
Factors controlling design of noncatalytic fluidsolid reactors:
 Reaction kinetics for single particles, e.g. eqn.(1) or eqn.(10).
 Size distribution of solids
 Flow patterns of solids and fluids in reactor (see figure below)
In systems where the kinetics are complex and not well known, or the products of reaction form a
blanketing fluid phase, or large temperature variations exist from position to position:
Analysis difficult design based on experience
(E.g. Blast furnace for producing iron)
However, some real systems can be adequately approximated by idealised systems. Ideal models can
also be used as a starting point (preliminary design) of complex systems.
Flow patterns for fluidsolid reactors:
(ref. Levenspiel (1999))
(What are the flow patterns in the above reactors?)
Page 85 of 90
 Consider two idealised systems:
(i) Plug flow of solid; uniform and constant gas composition; particles of unchanging but
different size.
(ii) As for (i) but solids in mixed flow.
6.3.1 Plug Flow of Solids
 F
[Ri]
volumetric flow rate of material of radius ~ R
i
fed to the reactor
i.e. Total feed rate, F:
=
max
0
] [
R
R
i
F F cm
3
/s
 Mean conversion of solid material,




.

\




.

\




.

\

i
R
i i
R size of
fraction
feed
R size of
particles in
d unconverte B
i.e.


.

\

=
max
0
] [
] [
) 1 ( 1
R
R
R B B
F
F
X X
i
i
(14)
Note: For R
i
<
'
i
R , where
'
i
R is the radius of the largest particle completely converted in a reactor
of spacetime t, then X
B
= 1.
E.g.
Reconsider the previous example on catalyst decoking of 2mm pellets (R=1000 m) in 20s. What
is
B
x if the feed contains particles of the following size distribution:
30% ~ 750 m
30% ~ 1000 m
40% ~ 1250 m
and the reactor is operated with a spacetime of ~22s?
Solution:
For this case, there will be complete conversion of 750 and 1000 m particles.
) 4 . 0 )( 1 ( ) 3 . 0 )( 1 1 ( ) 3 . 0 )( 1 1 ( ) 1 (
] 1250 [ B B
X X + + =
How do we calculate X
B
?
B B
B B B B
B
f R
f R f R
X
t
t t
3
0
3 3
0
3
4
3
4
3
4
=
3
0
3 3
0
R
R R
X
B
=
where R is the radius of the pellets after 22s in the reactor.
Page 86 of 90
Using eqn.(12) we can solve for R at t = 22s:
R ~ 440 m
956 . 0
1250
440 1250
3
3 3
] 1250 [
=
=
B
X
982 . 0 4 . 0 ) 956 . 0 1 ( 1 = =
B
X
What if the given size distribution data is continuous rather than discrete?
What if C
A
in gas phase is not constant?
6.3.2 Mixed Flow of Solids
 E.g. Fluidised bed reactor
 X
B
= f(residence time of solid)
From Reaction Engineering I, exitage distribution for material in a CSTR is given by:
t
t
t
e
t E
= ) ( (15)
Again, eqn.(14) is applicable, but this time, for a given particle size R
i
, there will be a distribution
of X
B
due to the distribution of residence time given by eqn.(15).
Thus:


.

\

=
max
0
] [
'
] [
) 1 ( 1
R
R
R B B
F
F
X X
i
i
(16)
where
'
B
X is the mean conversion of particles of size R
i
in the bed, i.e.
dt t E X X
i
R B B
)] ( ) 1 [( 1
0
] [
'
=
}
(17)
and X
B
is again attained from a suitable kinetic model, such as eqn.(12).
Note: in eqn.(17) corresponds to the time for the complete conversion (e.g. regeneration) of a
particle (core) of original radius R
i
.
Page 87 of 90
7 Notation
a, b, ... Stoichiometric coefficients 
a Area available for mass flux; crosssectional area m
2
a
c
Specific surface area of pellet on a reactor volume basis m
1
a
m
Specific surface area of pellet on a mass basis m
2
kg
1
a
m
Specific surface area of pellet on a bedvolume basis m
2
m
3
a
v
Specific surface area of reactor on a reactor volume basis;
specific surface area of gasliquid interface on a liquid volume
basis
m
1
a
v
Specific surface area of gasliquid interface on a reactor
volume basis
m
1
A Bed crosssectional area m
2
A
p
External surface area of pellet (chapter 2); projected particle
area (chapter 4)
m
2
Bi
m
Biot number for mass transfer 
C Total fluid phase concentration mol m
3
C
A
Bulk fluid phase concentration of A mol m
3
C
A0
Initial feed / bulk fluid phase concentration of A mol m
3
C
Ai
Liquid phase concentration of A at gasliquid interface mol m
3
C
As
Fluid phase concentration of A within catalyst mol m
3
C
S
As
Fluid phase concentration of A at catalyst / solid surface mol m
3
C
Ab
Fluid phase concentration of A within bubbles mol m
3
C
Ac
Fluid phase concentration of A within bubble cloud mol m
3
C
Ae
Fluid phase concentration of A within emulsion mol m
3
C
Coke concentration kg
coke
kg
catalyst
1
C
D
Drag coefficient 
C
p
Fluid heat capacity (molar basis) J mol
1
K
1
C
pg
, C
ps
Fluid / solid heat capacity (mass basis) J kg
1
K
1
d
p
Particle diameter m
d
*
p
Particle diameter for equal kinetic and mass transfer
resistances
m
d
pore
Pore diameter m
d
t
Tube / vessel diameter m
D
A
, D
B
Molecular diffusion coefficient m
2
s
1
D
AB
, D
m
Molecular diffusion coefficient m
2
s
1
D
e
Effective diffusion coefficient m
2
s
1
D
k
Knudsen diffusion coefficient m
2
s
1
D
p
Pore diffusion coefficient m
2
s
1
D
z
Axial dispersion coefficient m
2
s
1
E, E
ins
Enhancement factor; instantaneous reaction enhancement
factor

E Activation energy J mol
1
K
1
E(t) Exitage distribution 
E
1
, E
2
Ergun equation coefficients 
f
b
Fraction of bed occupied by bubbles m
3
bubbles
m
3
bed
f
B
Volume of B (e.g. coke) per unit pellet volume 
f
c
Cloud fraction in the bed m
3
cloud
m
3
bed
f
e
Fraction of bed occupied by emulsion gas m
3
eg
m
3
bed
F Force; volumetric flow rate of solid material (chapter 6) N; m
3
s
1
F
D
Drag force N
g Acceleration due to gravity m
2
s
1
Page 88 of 90
g
b
Mass of solids in bubble phase kg m
3
bubble
g
c
Mass of solids in cloud phase kg m
3
cloud
g
e
Mass of solids in emulsion phase kg m
3
emulsion
g
L
Mass of solids per unit volume of liquid kg m
3
liquid
G Mass flux; gas molar flow rate kg m
2
s
1
; mol s
1
Ga Galileo number 
j
D
, j
H
jfactors for mass and heat transfer, respectively 
h
f
Heat transfer coefficient J m
2
s
1
K
1
H Henrys law constant Pa m
3
mol
1
I Bessel function 
k Reaction rate constant mol
1n
m
3(n1)
s
1
k
G
Gasfilm mass transfer coefficient mol m
2
s
1
Pa
1
k
mc
Mass transfer coefficient based on concentration driving force m
3
f
m
2
s
s
1
k
mp
Mass transfer coefficient based on pressure driving force mol m
2
s
Pa
1
s
1
k
my
Mass transfer coefficient based on mole fraction driving force mol m
2
s
s
1
k
o
Overall reaction rate constant mol
1n
m
3(n1)
s
1
k
o
m
Mass transfer coefficient for equimolar counterdiffusion
conditions
k
s
Rate constant for a reaction rate based on per unit surface area m
3
f
m
2
s
s
1
(for n=1)
k
v
Rate constant for a reaction rate based on per unit volume s
1
(for n=1)
k
z
Thermal dispersion coefficient J m
1
K
1
s
1
K
I
Gas interchange coefficient between bubble and emulsion
phases
m
3
g
m
3
bed
s
1
K
Ib
Gas interchange coefficient between bubble and cloud phases m
3
g
m
3
bed
s
1
K
Ie
Gas interchange coefficient between cloud and emulsion
phases
m
3
g
m
3
bed
s
1
l Film or slab length m
L Reactor length; liquid molar flow rate m; mol s
1
m Total solids mass kg
g
Mass flow rate of gas kg s
1
s
Mass flow rate of solid kg s
1
M
m
Molecular weight g mol
1
n Reaction order 
n
A
Molar flow rate of A mol s
1
n
A0
Feed molar flow rate of A mol s
1
N
A
Molar flux of A mol m
2
s
1
N
c
Number of components 
N
T
Total molar flux mol m
2
s
1
Nu Nusselt number 
P Total pressure Pa
p
A
Partial pressure of A Pa
p
Ai
Partial pressure of A at gasliquid interface Pa
P
As
Partial pressure of A within catalyst Pa
P
s
As
Partial pressure of A at catalyst / solid interface Pa
Pe Peclet number 
Pr Prandtl number 
Q Heat flux J m
2
s
1
r Position coordinate within the pellet m
r
A
Rate of consumption of A mol m
2
s
s
1
; mol m
3
s
1
r
As
Rate of consumption of A within catalyst mol m
2
s
s
1
r
s
As
Rate of consumption of A at catalyst / solid surface mol m
2
s
s
1
r
A
Rate of consumption of A mol kg
1
s
s
1
Page 89 of 90
c
Rate of coke formation kg
coke
kg
catalyst
1
s
1
r
p
Pellet radius m
R Universal gas constant; radius of particle core J mol
1
K
1
; m
R
A
Effective reaction rate based on resistance in series model mol m
3
s
1
Re, Re
p
Reynolds number; Re based on particle diameter 
Sc Schmidt number 
Sh Sherwood number 
t Time S
T Temperature K
T
c
Coolant temperature K
T
s
Temperature within catalyst K
T
s
s
Temperature at catalyst / solid surface K
u Superficial fluid velocity m s
1
u
b
Bubble velocity m s
1
u
br
Bubble velocity under nonflow conditions m s
1
u
e
Emulsion gas velocity (interstitial) m s
1
u Superficial fluid velocity m s
1
u
mf
Minimum fluidisation velocity (superficial) m s
1
u
0
Superficial feed fluid velocity m s
1
u
t
Terminal velocity m s
1
U Overall heat transfer coefficient J m
2
s
1
v
0
Feed volumetric flowrate m
3
s
1
V Reactor volume m
3
V
core
Volume of particle core m
3
V
L
Liquid volume m
3
V
L
Liquid volume neglecting liquid film m
3
V
p
Pellet volume m
3
V Reactor volume m
3
x Liquid phase mole fraction 
X Degree of conversion 
y Length coordinate in liquid film m
y
A
Mole fraction of A 
y
As
Mole fraction of A within catalyst 
y
s
As
Mole fraction of A at catalyst / solid interface 
y
fa
Film factor 
y
L
Depth of liquid film m
z Length coordinate m
Greek
o
Constant
o
A
See equation (13c), chapter 2 
AH
r
Heat of reaction J mol
1
c
Voidage 
c
b
Bed void fraction 
c
g
Gas fraction per unit volume of liquid and gas 
c
p
Intraparticle void fraction 

Thiele modulus 
u
WeiszPrater modulus 
u
A
Catalyst deactivation factor based on the reaction of A 
Hatta number 
q
Intraparticle effectiveness factor 
Page 90 of 90
q
G
Global (intraparticle + external film) effectiveness factor 
v
Power law coefficient 
Gas viscosity kg m
1
s
1
= Pa s
Fluid thermal conductivity; mean free path W m
1
K
1
; m
b
Bulk density kg m
3
bed
B
Moles of B (e.g. coke) per unit volume of pellet mol m
3
pellet
f
Fluid density kg m
3
fluid
g
Gas density kg m
3
gas
p
,
s
Pellet/ solid density kg m
3
pellet
t
Space / residence time s
t
p
Tortuosity factor 
Subscript
0 Inlet/initial condition
b Bubble phase
c Cloud phase; coke
e, eg Emulsion phase; emulsion phase gas
f Fluid phase; fluidised bed
g Gas phase
G, L Gas / liquid phase
i Component identifier; interface
in, out Inlet / outlet streams
mf Minimum fluidisation
p Pellet phase
s Solid; surface
Superscript
s Pellet surface