1 Matlab solution to diusion-reaction problems

Diusion-Reaction problems are very common in chemical reaction engineering and often numerical
solutions are needed. Here we look at using matlab to obtain such solutions and get results of design
interest. Consider a model problem represented as:
d
2
c
dx
2
= f(c) (1)
which is a dimensionless form of the diusion with reaction problem. Here f(c) is a measure of the
reaction rate; For example, f(c) =
2
c for a rst order reaction where is the Thiele modulus. The
variable x is a dimensionless distance along the pore. The point x = 0 is taken as the pore mouth
and x = 1 pore end. The boundary conditions are taken as follows:
At x = 0, the dimensionless concentration, c = 1.
At x = 1, the gradient of the conncentration, dc/dx = 0.
We use the matlab program bvp4c to solve this problem. This requires that the Eqn. (1) be
written as two rst order equations rather than as a single second order dierential equation. This
can be done as follows: Consider a solution vector y with components y
1
and y
2
dened as follows:
y
1
= c and y
2
= dc/dx (2)
Eqn. (1) is then equivalent to the following two rst order equations.
dy
1
dx
= y
2
(= dc/dx) (3)
and
dy
2
dx
= f(y
1
) = (d
2
c/dx
2
) (4)
The boundary conditions are as follows:
at x= 0; y
1
= 1 Pore mouth (5)
and
at x = 1 y
2
= 0 Pore end (6)
The solution for the concentration prole and also for the local values of the gradient can then
be obtained using the function bvp4d in matlab. The eectiveness factor is related to the gradient
at the pore mouth and is calculated as:
= y
2
(0)/
2
(7)
The sample code for solving this problem is as follows:
---------------------------------------------------------------------------------------
% program calculates the concentration profiles and the effectiveness
% factor for m-th order reaction in a slab geometry.
global phi m phi2 n
global eta
% user actions, Specify parameters, a guess function for trial solution, a
% function odes to define the set of first order differential equations
1
% and a function bcs to specify the boundary conditions.
phi = 2.0 ; % thiele modulus for first reaction
m =2.0 ; % order of reaction
nmesh = 21 % intial mesh
nplot = nmesh ; % meshes for plotting the result.
% solution block.
x = linspace ( 0, 1, nmesh ) ;
solinit =bvpinit ( x , @guess ) % trial solution given by guess function
sol = bvp4c (@odes, @bcs, solinit) % bvp solved,
% post processing. plot concentration profiles, find eta
x = linspace ( 0, 1, nplot) ;
y = deval(sol, x) ;
y(1,:) % concentration profiles displayed
plot ( x, y(1,:) ) ; % these are plotted.
eta = -y(2,1) / phi^2
%______________________________________________________
function yinit = guess (x)
% provides a trial solution to start off
global phi m phi2 n
y1= exp(-phi*x)
y2= 0. * y1
yinit = [y1
y2 ] ;
%-------------------------------------------------------
function dydx = odes ( x, y )
% defines the rhs of set of first order differential equations
global phi m phi2 n
dydx = [ y(2)
phi^2 * y(1)^m
] ;
%-----------------------------------------------------------
function res = bcs ( ya , yb)
% provides the boundary conditions at the end points a and b
res = [ ya(1) - 1
yb(2) ] ;
------------------------------------------------------------------------
It is fairly easy to extend the code to multiple reactions. As an example consider a series reaction
represented as:
A B C
The governing equations are as follows assuming both reactions to be rst order.
d
2
c
A
dx
2
=
2
1
c
A
(8)
d
2
c
B
dx
2
=
2
1
c
A
+
2
2
c
B
(9)
2
The boundary condition at x = 0 (pore mouth) depend on the bulk concentrations of A and B. The
boundary condition at x = 1 (pore end) is the no ux condition for both A and B.
The solution vector y has size of four and consists of:
y =
_
_
_
_
y
1
= c
A
y
2
= dc
A
/dx
y
3
= c
B
y
4
= dc
B
/dx
_
_
_
_
(10)
The system is now formulated as four rst order ODEs for the four components of the solution
vector and solved by bvp4c in exactly the same way. Details are left out as homework problem.
An important consideration in this type of problem is how the selectivity is aected by pore
diusion. The yield of B is dened as:
Yield = ux of B out of the pore mouth / ux of A into the pore
which can be stated as:
Yield = (dc
b
/dx)
x=0
/(dc
a
/dx)
x=0
(11)
The relative concentration gradients of A and B at the pore mouth deterimines the local yield of B.
A smaller catalyst size is favrobale to B production. A larger paricle traps B in the interior of the
pores allowing it to react further and form C. This reduces the yield.
Maximum yield = (k
1
C
A0
k
2
C
B0
)/k
1
C
A0
which is also equal to
maximum yield = 1
k
2
C
B0
k
1
C
A0
Pore diusion resistance causes the yield to decrease from the above maximum value. Maximum
yield is realized only at low values of the Thiele parameters.
Example simulated with matlab for a particular case illustartes this point. Also c
a0
= 1 and
c
b0
= 0 which may correspong to the inlet of the reactor (with no recycle).
Consider
1
= 2 and
2
= 1. Yield is found as 0.8067.
Now let the catalyst size be reduced to one half the original size. Then both
1
and
2
decrease.
The new value of yield is found to be 0.8764. Maximum yield is 1 for this case. The pore diusion
resistance has decreased the yield. Students should verify these using matlab.
2 Solution For gas-liquid reactions
A(g > l) + B(l) Products
Governing equations are as follows:
D
A
d
2
C
A
dx
2
= k
2
C
A
C
B
(12)
D
B
d
2
C
B
dx
2
= k
2
C
A
C
B
(13)
Here x is the acutal distance into the im with x = 0 representing the gas-liquid interface.
We now introduce the following dimensionless parameters.
3
a = C
A
/C

A
where C

A
is the equilibrium solubility of gas A in the liquid corresponding to the partial pressure
of A in th bulk gas.
C

A
= p
gA
/H
A
where H
A
is the Henrys law constant for the species A.
b = C
B
/C
Bl
where c
Bl
is bulk liquid concentration of B.
Finally let be the dimensional distance in the lm (= x/). With these variables. the governing
equations are the following:
d
2
a
d
2
= Ha
2
ab (14)
d
2
a
d
2
= Ha
2
ab/q (15)
where HA and q are two dimensionless quantities.
Ha
2
=
2
k
2
C
Bl
D
A
and
q =
D
B
C
BL
D
A
C

A
Noting that k
L
= D
A
/, the Hatta number can also be expressed as:
Ha
2
=
D
A
k
2
C
Bl
k
2
L
(16)
The boundary conditions for species A are as follows:
Case 1: No gas im resistance:
At = 0, C
A
= C

A
and Hence a = 1.
At = 1, C
A
= C
AL
is some specied value depending on the bulk processes. In other words,
C
AL
will depend on the extent of bulk reactions, convective and dispersive ow into the bulk etc.
But even for moderately fast reactions, the bulk concentration of dissolved gas turns out be zero and
we take this value to be zero. Hence a = 0 at = 1 will be used as the second boudnary condition.
The validity of this assumption can be checked after a solution is obtained and modied if needed.
Case 2: Gas lm resistance included. A balance over the gas lm provides the boundary condition:
k
g
(p
gA
p
iA
) = D
A
dC
A
dx
(17)
Also note that p
iA
the interfacial partial pressure of A is related to the interfacial concentration of
A in the liquid by Henrys law. Thus C
A
(x = 0) = P
ia
/H. Using these the boundary condition can
be expressed in dimensionless form as
Bi
g
(1 a
=0
) =
_
da
d
_
=0
(18)
4
where Bi
G
= k
g
H/k
L
, A biot number for gas side mass transfer. The boundary condition is now
of the Robin type. The boundary conditon at = 1 is the same as before. (take as a = 0 if the
reactions are reasonably fast. Otherwise a need a mass balance for A in the bulk liquid as well.)
The boundary condition for species B are specied as follows:
At = 0 we use db/d = 0 since B is non-volatile and the ux is therefore zero.
At = 1, we have b = 1. This completes the problem denition. In order to solve this using
matlab, the governing (two) equations are cast as four rst order dierential equation. The solution
vector y is now given by:
y =
_
_
_
_
y
1
= a
y
2
= da/d
y
3
= b
y
4
= db/d
_
_
_
_
(19)
The matlab program to do these calculations is shown below. The program structure is similar
to Thiele. Note that A and B are counterdiusing in this case and the boundary condition section
reects these changes. The program can be used to calculate the rate of absorption or to learn how
the regime of absorption changes with dimensionless parameters. Thus the program is an eective
learning tool. In the following sections, these features are illustrated.
function hatta
% program calculates the concentration profiles for gas A
% and liquid reactant B in the film.
% 11-1-03 P. A. Ramachandran
global m q bi_gas
% The dimensionless parameters are calculated in a separate function.
% m = Hatta number, q = q parameter, bi_Gas = Biot number for gas.
t = 1.0 ; % a dummy parameter
[ m , q , bi_gas ] = calculate_these ( t ) ;
% numerical solution.
nmesh = 21 % intial mesh
nplot = nmesh ; % meshes for plotting the result.
% solution block.
x = linspace ( 0, 1, nmesh ) ;
solinit =bvpinit ( x , @guess ) % trial solution generated by guess function
sol = bvp4c (@odes, @bcs, solinit) % bvp solved,
% post processing. plot concentration profiles, find eta
x = linspace ( 0, 1, nplot) ;
y = deval(sol, x) ;
y(1,:) % concentration profiles displayed
y(3,:) % concentration profile of B
plot ( x, y(1,:) ) ; hold on ; plot (x, y(3,:) ) % these are plotted.
EnhanceFactor = -y(2,1)
b_interface = y(3,1)
bi = 1. + 1/q - EnhanceFactor /q
%______________________________________________________
function yinit = guess (x)
global m q bi_gas
5
y1= exp(-m*x)
y2= 0. * y1
yinit = [y1
y2
y2
y2] ; % some arbitrary guess values here
%-------------------------------------------------------
function dydx = odes ( x, y )
global m q bi_gas
dydx = [ y(2)
(m^2 * y(1)* y(3) )
y(4)
(m^2 * y(1)* y(3) )/q ] ;
%-----------------------------------------------------------
function res = bcs ( ya , yb)
% This is set up for na gas film resistance.
res = [ (ya(1) - 1)
(ya(4))
(yb(1) )
(yb(3) -1) ] ;
%_______________________________________________________________
function [ m , q , bi_gas ] = calculate_these ( t )
da = 1.40e-09 ;
k2= 10.02;
db = 0.77e-09 ;
kl = 2.2e-04;
b0 = 1000. ;
astar = 6.25 ;
msq = da* k2 * b0 / kl^2
m = msq^(0.5)
db = 0.77e-09;
z = 2. ;
q = db * b0 / ( z * da* astar )
bi_gas = 1000.0; % assigned here.
---------------------------------------------------------------------
A result of sample simulation xing q and varying Hatta is shown in Figure 1.
The eect of large Hatta is shown in gure 2 where the reaction approaches an instantaneous
asymptote.
E

= 1 + 1
Example; H
2
S Removal
6
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.5
1
Distance into film
C
o
n
c
e
n
t
r
a
t
io
n
,

a

o
r

b

M = 10 ; E = 6.43
M= 100; E = 9.9605
Results for q = 9
Figure 1: Concentration proles for dissolved gas A and liquid phase reactant B for two values of
Hatta number for a xed value of q. Note that the species B depletes more in the im for larger
value of M = Ha,
Lo-Cat is a process for removing H
2
S from renery ue gases and uses a chelated Fe(3+)EDTA
solution. A simple representation of the reaction is:
H
2
S + 2Fe
+3
2H
+
+ S
0
+ Fe
+2
Find the rate of absorption of H2S if the partial pressure in the gas phase is 0.05 atm and the
concentration of Fe
+++
in the liquid is 60 mol/m
3
.
The kinetic rate constant is 9m
3
/mol. s ( A second order reaction).
The diusion coecients in the liquid are:
D
A
= 1.44 10
9
m
2
/s
D
B
= 0.54 10
9
m
2
/s.
Henry law coecient is 1950 Pa/mol m
3
Liquid side mass transfer coecient, k
L
= 2 10
4
m/s,.
Solution:
The equilibrium concentration of H2S at the interface = C

A
= p
gA
/H = 2.56 mol/m
3
. The q
parameter is calculated as:
q =
D
B
C
BL
D
A
C

A
= 4.4
The Hatta number is calculated as:
Ha =

D
A
k
2
C
B
l
k
L
= 4.4
Running the Matlab program Hatta.m for these parameters gives the enhancement factor as 3.12.
The program also gives the complete concentration proles for H
2
S abnd Fe EDTA complex in
the lm.
Using the calculated enhancement factor, thh rate of absorption is calculated as;
Rate of absorption per unit interfacial area = k
L
C

A
E = 0.0016mol/m
2
s.
7
3 Asympototic Solution to diusion-reaction problems
For non-linear problems, a solution can be obtained analytically for large values of Thiele modulus.
This asympotic solution can then be used to dene a generalized Thiele modulus and can be extended
to obtain approximate analytical solutions even for low Thiele modulus. The procdure is illustated
below.
Consider a model problem represented as:
D
e
d
2
c
dx
2
= f(C) (20)
where f(C) is the rate of consumption of species. The order of the dierential equation can be
reduced by one by using the transformaiton p = dC/dx. Noting that
d
2
C
dx
2
= p
dp
dC
(21)
we nd that Eq. (20) can be written as
D
e
p
dp
dC
= f(C) (22)
The above equation can be solved by separation of variables. Two limits are needed to obtain
the value of the concentration gradient at the surface, p
s
. At the pore end, the value of the gradient
p is zero. Also for large values of Thiele modulus the concentration drops to nearly zero at some
interior point in the catalyst. hence the concentration can be set as zero at the pore end. Thus we
can set C = 0 at p =0 as one of the limits for integraion of Eqn. (22). Let the surface gradient be
desgnated as p
s
and the concentration at the surface as c
s
. This sets another limit for integration
of Eqn 21 which can be now be written as:
D
e
_
ps
0
pdp =
_
Cs
0
f(C)dC (23)
Integrating and rearranging
p
s
=
_
2/D
e
_
_
Cs
0
f(C)dC
__
1/2
(24)
The minus sign is used for the square root since the concentration is decreasing with increase in x.
The eectiveness factor is given by the ratio of the actual rate to that based on surface concentration.
Actual rate = D
e
p
s
S
p
(25)
by Ficks law applied at the surface. Hence the eectivess factor is gven by:
= D
e
p
s
S
p
/V
p
f(C
s
) (26)
Using the Eq. 24 for p
s
we nd
= (S
p
/V
p
)f(C
s
)
_
2D
e
_
Cs
0
f(C)dC
_
1/2
(27)
8
The above result is applicable for any kinetics provided we are in the asympotitc region. For a rst
order reaction = 1/ for large Thiele modulus. In order to make the results in the same format it
is convenient to dene a Thiele modulus for any kinetics as:
= (V
p
/S
p
)f(C
s
)
_
2D
e
_
Cs
0
f(C)dC
_
1/2
(28)
which is known as the generalized Thiele modulus. Now we have = 1/ for any kinetics but only
in the asymptotic case. The expression for is now generalized as as:
=
tanh()

(29)
The above expression is strictly valid only in the asymptotic region where tanh() tend one but is
used for the entire range of values as an approximate solution.
Problems on asymptotic solution
1. What form does the Thiele modulus take for an m-th reaction? What is the eect of sur-
face conventration of the Thiele modulus? What is the eect of surface concentration on
the eectiveness factor in the asymptotic region for (a) second order reaction, (b) half order
reaction?
2. What form does the expression for the generalized Thiele modulus take for a L-H kinetics of
the form
f(C) = kC/(1 + KC)
Write an analytical expression for the eectiveness factor for such reactions. Express the result
in terms of the two dimensionless groups: (i) kL
2
/D
e
and (ii) KC
s
. Find the values of for
some chosen values of these parameters and compare it with the numerical solution generated
using matlab. Find the region where the errors are maximum. What is the order of maximum
error in the approximate analytical solution?
3. Zero order kinetics have special features. The eectivenes factor is equal to 1 up to a value of
Thiele modulus at which the concentration at the center of catalyst becomes zero. Find this
critical Thiele modulus. What are the operational implications of this result?
4 Eective Diusion Coecient
The pore diusion rate plays an important role in aecting the rate of gas-solid reactions as shown
in the earlier sections. The eective diusivity in the pore D
e
is however a medium propery and
some models are useful to estimate these in the absence of actual data. The diusion within the
catalyst proceed by two mechanisms: (1) Bulk diusion within the pores.(2) Bombardment with the
walls of the pore if the pore radius is small. The second mechanism is known as Knudsen diusion.
The bulk diusion within the pores is a result of gas-gas collisiton and for a single pore the
diusivity is the same as the gas phase binary diusivity (for ab inary pair). This value of diusivity
is corrected for the pore structure by incorporating two factors (1) porosity, which accounts for
the reduced area accessible for diusion and (2) tortuosity, which accounts for the non-straight
path for diusion.
D
b, eff
= D
12

9
Knudsen diusion occurs when the size is the pores of the order of mean free path of the diusivity
molecule. The molecules now collide more frequently with the pore walls rather than with other
molecules. the Knudsen diusion coecient in a cylindrical pore is given as:
D
K
=
2
3
r
e
V
1
where r
e
= is the eective pore radius and V
1
is the average molecular speed of species 1. This is
given from the kinetic theory of gases as:
V
1
=
_
8 RT
M
1
_
1/2
Substituting the values of gas constants, we nd the following dimensional equation for the
Knudsen diusion coecient for species 1:
D
K,1
= 97 r
e
_
T
M
1
_
1/2
(30)
in m
2
_
s with r
e
in m and T in Kelvins.
The Knudsen diusion coecient in the acutal pore is then corrected for the porosity and tor-
tuosity and is given by:
D
K, eff
=

D
K
The phenomena of ordinary diusion and Knudsen diusion, may be occurring simultaneously.
The two can be combined by the following formula:
1
D
1, eff
=
1
D
b, eff
+
1
D
K, eff
Additional mechanism is the surface diusion where the adsorbed species migrates along the surface.
The eects are often ignored if the pores are relatively large. This phenomena may be of importance
in monolith type of catalysts.
5 Diusion in bidisperse media
Catalysts often have a bimodal pore size distribution with micropores and macropores. The model
of Wakao-Smith is useful for predicting the eective diusion coecients in such systems. According
to their model, the eective diusivity is given as:
D
e
=
2
M
D
M
+
2

(1 + 3
M
)
(1
M
)
D

(31)
where
M
is the void fraction associated with the macropores and D
M
is the diusivity associated
with it. Similarly, the term

is the void fraction associated with the micropores The term, D

in
the above equation is the diusion coecient associated with the micropores. The diusivities are
calculated by combining the molecular and Knudsen contributions in parallel as:
D
M or
= D
m
D
K
/(D
m
+ D
K
) (32)
10
where D
m
is the molecular diusivity and D
K
is the Knudsen diusivity calculated according to Eqn.
(30). The macropore radius, r
M
and micropore radius, r

are needed for the Kundsen calculations.

/r

(33)
where S
in
is the internal surface area per unit catalyst volume. The above expression assumes that
the pores are cylindrical. The rate is generally proportional to the surface area. Thus if k
s
is the
rate constant per unit surface area, then the rate constant k
v
based on the catalyst volume is:
k
v
= k
s
S
in
(34)
The Thiele modulus for a rst order reaction dened earlier as;
= (V
p
/S
p
)(k
s
S
in
/D
e
)
1/2
) (35)
is thus a complex function of the pore structure. Qualitatively speaking, the micropores increase the
area but have lower diusivity. The macropores have the opposite eect. They are easily accessible
(larger diusivity) but contribute less to the rate constant since they have lower surface area per
volume. Hence there is an optimum catalyst structure for a given reaction that gives the best
eectiveness factor. The above considerations provide a simple basis to optimize the pore sturctures
of the catalysts.
6 Stefan-Maxwell model for multicomponent diusion
Mass transport in isothermal multicomponent systems in the gas phase is described by the following
constitutive equation.

dC
i
dx
=
n

j=1
y
j
N
i
y
i
N
j
D
ij
(36)
where n is the number of components and D
ij
is the binary pair diuvisity in the gas phase.
For porous catalysts the equation is modied by including both Knudsen and bulk diusion.

dC
i
dx
=
N
i
D
Ke,i
+
n

j=1
y
j
N
i
y
i
N
j
D
ij,e
(37)
The mass balance for a dierential control volume (for a 1-D slab geometry) gives
dN
i
dx
=
nr

ji
r
j
(38)
where the RHS is the total rate of production of species i from the various reactions. The term r
j
is
the rate function for the j-th reaction which is the function of the local concentrations of the various
species. Eqn. (37 and 38 have to be integrated together now.
Note that Eqn. 37 can be written in terms of the component partial pressures as:

1
R
G
T
dp
i
dx
=
N
i
D
Ke,i
+
n

j=1
p
j
N
i
p
i
N
j
P
t
D
ij,e
(39)
11
where P
t
is the total pressure and R
G
is the gas constant. Since the rates of reactions are usually
correlated as a function of partial pressures, the above form is often more convienent. The viscous
contribution to ow is neglected in these equations. It does not mean that the catalyst pellets are
isobaric; pressure gradients can exisit in the pellets and they are implicitly accounted for by the use
of the constitutive equations in terms of the partial pressures rather than in terms of component
mole fractions.
A sturucture of a matlab program for solving these equations is presented here.
function StefanMaxwell ()
% program calculates the concentration profiles and the effectiveness
% factor based on Stefan-Maxwell model.
% 3-07-05 written by P. A. Ramachandran
% test problem 1 in the paper by Haynes
% user actions, Specify parameters, a guess function for trial solution, a
% function odes to define the set of first order differential equations
% and a function bcs to specify the boundary conditions.
% also define rate in the function ratemodel.
global ns neq
ns = 4 % four species
neq = 8 % 4*2 equations
% Eight variables Y(1) to y(4) are the concentrations of CO, H2O, CO2 and
% H2, y(5) to y(8) are the flux values.
% parameters used are as follows:
global temp pressure
global ctot
temp = 400+273.;
pressure = 25.0 ; % atm
Rgas = 82.06 % gas contant in cm^3 atm /g-mole K
ctot = pressure/Rgas/temp; % total concentration mole/c.c
% surface (pore mouth) mole fraction
global ys
ys(1) = 0.1; ys(2) = 0.6; ys(3) = 0.08; ys(4) = 0.22 ;
ys = ys * ctot % surface molar concentrations.
% rate and other parameters
global k rhop Keq
k = 2.05e-04 ;
rhop = 1.84 ;
Keq = 12.0
% diffusion parameters; dk = knudsen ; db = binary pair (a matrix)
global dk db
dk(1) = 0.00649;
dk(2) = 0.00494 ;
dk(3) = 0.0098;
12
dk(4) = 0.0231;
% **** add db values for binary pairs
% catalyst properties
length = 0.1588 ;
rate_s = reactionrate(ys) ; % rate based on surface values
nmesh = 21 % intial mesh
nplot = nmesh ; % meshes for plotting the result.
% solution block.
x = linspace ( 0, 1, nmesh ) ;
solinit =bvpinit ( x , @guess ) % trial solution generated by guess function
sol = bvp4c (@odes, @bcs, solinit) % bvp solved,
% post processing. plot concentration profiles, find eta
x = linspace ( 0, length, nplot) ;
y = deval(sol, x) ;
% concentration profiles of various components.
y(1,:)
y(3,:) ;
y(2,:)
y(4,:) ;
plot ( x, y(1,:) , -+ ) ;% hold on ; plot (x, y(2,:), -+ ) % these are plotted.
% flux values for the four species.
y(5,1)
y(6,1)
y(7,1)
y(8,1)
rate_s
eta = y(5,1)/rate_s /0.1588
%______________________________________________________
function yinit = guess (x)
global phi m phi2 n
global ys
global k rhop Keq
% find rate based on surface values and use it as trial value for flux
%rate =rhop* k * ys(1)^0.9 * ys(2)^0.3 * ys(3)^(-0.6) * (1.- ys(3) * ys(4)/Keq/ys(1) /ys(2) )
rate= reactionrate(ys) ;
yinit = [ys(1)
ys(2)
ys(3)
ys(4)
-rate
-rate
rate
rate ] ; % guess values here
%-------------------------------------------------------
function dydx = odes ( x, y )
global ns neq
global ctot
13
global k rhop Keq
global dk db
rate = reactionrate(y);
% bulk diffusion terms to be added. Currently set as zero
bulkterm = zeros (ns,1);
dydx = [- y(5) /dk(1) + bulkterm(1)
- y(6)/dk(2)
-y(7)/dk(3)
-y(8)/dk(4)
-rate
-rate
rate
rate ] ;
%-----------------------------------------------------------
function res = bcs ( ya , yb)
global a0 b0
global ys
res = [ ya(1) - ys(1)
ya(2) - ys(2)
ya(3) - ys(3)
ya(4) - ys(4)
yb(5)
yb(6)
yb(7)
yb(8) ] ;
%___________________________________________________________________
function ratemodel = reactionrate(y)
global k rhop Keq
global temp pressure
global ctot
p = y*82.06*temp % species partial pressures
ratemodel =rhop* k * p(1)^0.9 * p(2)^0.3 * p(3)^(-0.6) * (1.- p(3) * p(4)/Keq/p(1) /p(2) ) ;
14
7 Temperature Eects
The simultaneous transport of heat and mass in a porous catalyst can be represented by the following
set of equations.
D
e
d
2
C
dx
2
= f(C, T) (40)
k
e
d
2
T
dx
2
= (H)f(C, T) (41)
where k
e
is the eective conductivity of the catalyst and H is the heat of reaction. The local rate
(of disapperance) f(C, T) is now a function of both the pore concentration and the temperature at
any point in the pellet.
The boundary conditions are specied as follows: No ux condition is imposed at the pore end
(x = 0) for both heat and mass.
at x =0 dC/dx = 0; dT/dx = 0
At the pore mouth (x = L) the ux into the pellet must match the transport through the gas lm.
This leads to the following boundary conditions.
For mass transport
D
e
_
dT
dx
_
x=L
= k
m
(C
g
C
x=L
)
For heat transport
k
e
_
dT
dx
_
x=L
= h(T
g
T
x=L
)
The equations can be put in terms of dimensionless variables by scaling the concentration and
temperautre by reference values. The bulk concentration and temperature can be used as the
reference values if the particle scale model is being analyzed. For reactor scale models, the inlet feed
conditions can be used as reference.
The dimensionless mass and heat transport equations are then as follows:
d
2
c
d
2
=
f(C
ref
, T
ref
)L
2
D
e
C
ref
[f

(c, )]) (42)

is a dimensionless rate dened as:

= f(C, T)/[f(C
ref
, T
ref
)]
The rst term on the RHS of 42 can be identied as a Thiele parameter
Let
2
=
f(C
ref
, T
ref
)L
2
D
e
C
ref
Then the mass balance equation is:
d
2
c
d
2
=
2
f

(c, ) (43)
The heat balance equation can be represneted as:
d
2

d
2
=
2
f

(c, ) (44)
15
where is the dimensionless parameter (thermicity group) dened as:
=
(H)D
e
C
ref
k
e
T
ref
(45)
Note thet takes a positive value for exothermic reactions and a negative value for endothermic
reactions.
The boundary conditions can be normalized as:
at = 0 dc/d = 0; d/d = 0
and the conditions at = 1 can be related to the Biot numbers for heat and mass
_
dc
d
_
s
= Bi
m
(c
g
c
s
))
_
d
d
_
s
= Bi
h
(
g

s
))
where the subscript s is used for the = 1 i.e,. the surface values. Here we dene
Bi
m
= k
m
L/D
e
and
Bi
h
= hL/k
e
The magnitude of the internal temperature gradient can be assessed by combining the above two
equations without even actually solving the equations. The mass and heat transport equations 54
and 55 can be combined as:
d
2

d
2
+
d
2
c
d
2
(46)
Hence the integration of the above equation twice leads to

s
= (c
s
c) (47)
The maximum value of the temperature occurs for an exothermic reaction when the center concen-
tration drops to zero. This occurs in the strong pore diusion resistance for mass transfer. Hence
an estimate of the internal gradients is:
Max
c

s
= c
s
= O() (48)
where
c
is the center tempearure and the symbol O means of the order of.
Also the magnitudes of the external gradients can be estimated by the following procedure: Here
we introduce the overall eectiveness factor
0
dened as:
0
= actual rate / rate based on reference
conditions
Rate of transport to the surface is equal to S
p
k
m
(C
g
C
s
)
Rate of reaction is V
p
f(C
ref
, T
ref
)
0
.
Equating the two and using the dimensionless groups we nd as estimate of the external concen-
tration gradients.
c
g
c
s
=

2

0
Bi
m
(49)
16
Note that
2

0
is the same as the Weisz modulus and can be directly calculated if the measured
data is available. The value of
0
need not be explciitly evaluted in order to use the above equation
for diagnostic purposes.

0
= (R
measured
)L
2
/(D
e
C
ref
) (50)
Similar analysis equating the heat transport in the lm to the total heat released from the
reaction leads to an estimate the lm temperararute dierence.

0
=

2

0
Bi
h
(51)
The above relations enables a rapid determination or the relative importance of external and
internal temperature gradients. It can be, for instance, used to determine if the pellet is operating
under near isothermal conditions or whether a temperature correction to the data is necessary.
The temperature rise in the pellet can be assessed as follows: Combining 49 and 51 we nd:

s
= c
g

_
1

2

0
Bi
m
_
(52)
which is a closer estimate of the pellet temperature rise than the earlier equation based on the
surface concentration.
*************************************************
transport coecient correlations
8 Numerical Solutions
Numerical solution for a rst order reaction is indicated in this section. The dimensionless rate for
a rst order reaction can be expressed as:
f

= c exp[(1 1/)] (53)

d
2
=
2
c exp[(1 1/) (55)
The boundary conditions are:
at = 0 dc/d = 0; d/d = 0
and = 1
_
dc
d
_
s
= Bi
m
(c
g
c
s
))
_
d
d
_
s
= Bi
h
(
g

s
))
17
For the particle problem (local particle scale model) we set c
g
and
g
as one.
We nd that even for this simple rst order reaction, the overall eectiveness factor is a function
of ve parameters: , , , Bi
m
and Bi
h
.
An illustrative plot of the eectivess factor vs Thiele modulus is shown in Fig 1 based on the
work of Aris and Hateld. It is clear that the behavior is complex and multiple steady states are
observered. For one case in the g we see two distinct regions of multiple solutions. The region at
low Thiele modulus is caused by intraparticle gradients while the second one is due to interparticle
gradients. The two regions may overlap producing ve steady states as seen in the second case
shown in Fig 1.
Sample result was simulated using the matlab program for the following value of the parameters.
= 0.5, = 1/3, = 27, Bi
m
= 300 and Bi
h
= 100.
Model for Isothermal Pellet
An useful simplication is to assume that there are no internal gradients for temperature. The
temperature changes are conned only to the gas lm. This simplies the calculations since the only
the intraparticle concentration proles need to be solved. Approximate analytical solutions shown
earlier can be used for the internal eectiveness factor. The particle temperature is not known and
has to be solved iteratively. The calculation procudre is indicated below for a rst order reaction.
The Thiele modulus is dened on the basis of the surface temperature
= exp[/2.(1 1/
s
)] (56)
where is the Thiele modulus based on refernce conditions. The overall eectiveness factor for a
rst order reaction was shown earlier as:

0
=
The surface temperature is related to the gas phase temperature by the equation shown earlier which
is reproduced below for completeness.

s
=
0
+

2

0
Bi
h
(57)
The above equations are solved simultaneously or by trial and error to obtain the overall eectivess
factor and the surface temperature.
Example : Calculate the eectiveness factor for the following conditions;
= 2., = 0.05, = 20, Bi
m
= 200 and Bi
h
= 20.
Solution: Since is small let us assume that all the temperature drop in conned to the gas lm.
To start o the calculations assume a particle temperature of 1.105Then the following set of results
can be obtained.
=

c
=

0
The new value of the surface temperature is calculated using 57 as:

s
=
18
Problems on Pore diusion
1. Calculate the diusion coecient for a catalyst with the following properties based on the
Wakao-Smith model
This system is H
2
Sair. Find the Thiele modulus and eectiveness factor if the rate constant
is :***
Find the eect of changing it to a mesoporous support where the pore diameter is 6 to 10 nm.
Find the eect of changing the macropore radius.
19
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.5
1
Distance in to the film
C
o
n
c
e
n
t
r
a
t
io
n

o
f

A

a
n
d

B

M = 1000; q = 9.
E = 10.;
= 0.1 ;
Figure 2: Concentration proles for dissolved gas A and liquid phase reactant B for a large values
of Hatta number for a xed value of q. The reaction is now instantaneous; Note that the reaction is
now conned to a narrow zone.
20