Contents

1 Mass transfer effects in reaction engineering 1

1.1 Internal effectiveness factors . . . . . . . . . . . . . . . . . . . 1
1.1.1 A worked example . . . . . . . . . . . . . . . . . . . . 6

2 The effectiveness factor in action 9

3 Summary 10
<h1>Mass transfer in porous catalysts (Ch 12 in Fogler)</h1>
https://img-9gag-fun.9cache.com/photo/287396_700b.jpg Mass trans-
fer

1 Mass transfer effects in reaction engineering

• In this section we consider the effects of mass transfer, e.g. diffusion,
on reactor design

• The general strategy is to derive a correction factor that simply mod-

ifies the rate law

• This enables us to use the same design equations

1.1 Internal effectiveness factors

• When porous catalyst pellets are used, reactants must diffuse into the
pellet to react, and then products must diffuse out of pellet to exit the
reactor.

1
• If there is resistance to diffusion, e.g. the pellets are large, or with
small pores, then we must consider this effect on the reactivity of the
pellet

• the effective diffusion coefficient is defined as:

DAB φp σc
De =
τ
• where DAB is the normal gas phase diffusivity

• φp is the pellet porosity (void volume / total volume)

• σc is the constriction factor, which is ratio of the smallest radius in the

pores to the largest radius

2
• τ is the tortuosity factor, which is the actual distance traveled by a
molecule in the particle divided by the shortest distance it could have
traveled.

• all of these factors tend to make De smaller than DAB

#+ipynb-newcell

• When there are diffusion limits, the concentration inside the pellet is
not the same as the concentration outside the pellet because there are
reactions occuring inside the pellet

• The concentration can be similar to the outside concentration if the

reaction rate is slow and there is no diffusion limit

• It may be smaller, or practically zero inside the pellet if the rate of

reaction is very fast compared to diffusion

• We would like to calculate the effective rate inside the particle, because
then we could determine its effect on reactor design

3
• Since the effective rate will be less than the ideal rate, the result is we
will need larger reactors because the particles are not effectively used

– We will see later it is sometimes possible to design better particles

• To quantitatively determine the significance, we consider a mole bal-

ance on a spherical particle.

#+ipynb-newcell

• Let WA be the molar flux of A

• We have at r the rate of A into the shell as WA,r 4πr2 |r

• We have the rate of generation of A in the shell as rA

0 ρ 4πr 2 ∆r
c m

• In the limit of ∆r → 0 and at steady state we have

dWA,r r2 0 ρ r2 = 0
dr − rA c

• For dilute concentrations (typical of gas phase reactions)

4
 – WA,r = −De dC
dr
A

• Let us assume the rate law is simple, e.g. rA = kn CA

n

• This leads to
d2 CA 2 dCA kn n
dr2
+ r dr − D e CA =0

• You can see that reaction and diffusion are coupled

• The rate of diffusion is driven by gradients in concentration

• The rate of reaction is driven by concentration

• This is not an ordinary differential equation!

• We have boundary conditions

– CA = CA,s at r = R
– dCA
dr = 0 at r = 0
– This is a second order boundary value problem

• To solve the problem, we convert it to a system of first order differential

equations

• Let WA = dCA
dr

d2 CA
– then dWA
dr = dr2

Finally, we have these two coupled equations in the standard form:

dWA 2 kn n
= − WA + C (1)
dr r De A
dCA
= WA (2)
dr
With these new conditions: CA (R) = CA,s and dCA
dr |r=0 =0

• If we solve these equations, we will have a function that tells us the

concentration of A inside the particle.

– Then we can integrate the reaction rate throughout the particle

to determine the effective rate (which will be less than the rate if
the concentration is uniform throughout the particle)

5
 RR
– The overall rate is 0 kn CAn 4πr 2 dr
RR
– The ideal rate is 0 kn CA,s
n 4πr 2 dr

– We define the effectiveness factor η = overallrate

idealrate
– Finally, in a mole balance, we replace the ideal rate with the
effective rate: ref f = ηr

1.1.1 A worked example

Let us consider a particle with these properties"

• De = 0.1 cm2 /s

• R = 0.5 cm

• k = 6.4 1/s

• CA,s = 0.2 mol/L

• Our goal is to determine the effectiveness factor.

• Solution: This is a boundary value problem.

• Here we solve the problem using the shooting method. We know the
value of WA at r = 0: it is 0 because there is no flux about the center
of the sphere due to symmetry.

• We do not know, however, what the concentration is at r = 0. So, we

guess what the value is, and then integrate the differential equations
from r = 0 to r = R, and see if CA (R) = CA,s .

• If it does, we made a good guess. If it does not, we try a new guess

for CA (r = 0), and iterate until we find the answer. Here is code
that implements this idea. You cannot see the iteration in this code
because it is done by hand. Start with Ca0 = 0.05, and work down to
a solution.

import numpy as np
from scipy.integrate import odeint

%matplotlib inline
import matplotlib.pyplot as plt

6
De = 0.1 # diffusivity cm^2/s
R = 0.5 # particle radius, cm
k = 6.4 # rate constant (1/s)
CAs = 0.2 # concentration of A at outer radius of particle (mol/L)

def ode(Y, r):

Wa = Y[0] # molar rate of delivery of A to surface of particle
Ca = Y[1] # concentration of A in the particle at r

if r == 0:
dWadr = 0 # this solves the singularity at r = 0
else:
dWadr = -2*Wa/r + k/De*Ca

dCadr = Wa
return [dWadr, dCadr]

# Initial conditions
Ca0 = 0.029315 # Ca(0) (mol/L) guessed to satisfy Ca(R) = CAs
# It takes a lot of accuracy to get the solution
Wa0 = 0 # no flux at r=0 (mol/m^2/s)

rspan = np.linspace(0, R, 500)

Y = odeint(ode, [Wa0, Ca0], rspan)

Ca = Y[:, 1]

# here we check that Ca(R) = Cas

print(’At r={0} Ca={1}’.format(rspan[-1], Ca[-1]))

plt.plot(rspan, Ca)
plt.xlabel(’Particle radius’)
plt.ylabel(’$C_A$’)

r = rspan
eta_numerical = (np.trapz(k * Ca * 4 * np.pi * (r**2), r)
/ np.trapz(k * CAs * 4 * np.pi * (r**2), r))
print(’The effectiveness factor = ’, eta_numerical)

7
 You can see from the graphical solution that the concentration inside the
particle is much lower than outside the particle. As a result, the overall rate
of the particle is only about 56% of the ideal rate. Consequently, you would
need a larger weight of catalyst, or a larger reactor to achieve the same level
of conversion as if there were no mass transfer limitations.

• It is tedious to numerically solve for the effectiveness factor for each

situation

• Let us recast the equations in dimensionless form.

• Let the characteristic length scale be defined by the volume to surface

ratio: for a sphere this <div class="alert alert-warning"> a = R/3
</div>

• Now we define the dimensionless variables: r = r/a and c = CA /CAs .

• This leads to the dimensionless differential equation:

1 d 2 dc − Φ2 c = 0


r 2 dr r dr
2
or alternatively: dr
d c 2 dc 2
2 = − r dr + Φ c
with boundary conditions: c(r = 3) = 1
dr = 0 at r = 0
dc
q
ka2
where we have defined <div class="alert alert-warning"> Φ = DA
</div> which is known as the Thiele modulus, and it is a ratio of reaction
rate to diffusion rate.

• The point of this exercise is that an analytical solution to this problem

exists:
3 sinh Φr
c(r) = r sinh 3Φ

• One can then analytically calculate the effectiveness factor as the actual
rate of reaction in the particle divided by the ideal rate, to arrive at:

<div class="alert alert-warning"> η = Φ1 tanh1 3Φ − 3Φ 1

</div>
 

• This solution is plotted in two different ways below.

Phi = np.linspace(0, 100, 1000)

eta = 1.0 / Phi * (1.0 / np.tanh(3 * Phi) - 1.0 / (3 * Phi))

8
f, (ax1, ax2) = plt.subplots(1, 2)
ax1.plot(Phi, eta)
ax1.set_xlim([0, 20])
ax1.set_xlabel(r’$\Phi$’)
ax1.set_ylabel(r’$\eta$’)

ax2.loglog(Phi, eta)
ax2.loglog(Phi, 1.0 / Phi, ’k--’, label=r’1/$\Phi$’)
ax2.loglog(Phi, np.ones(shape=Phi.shape), ’b--’, label=’1’)
ax2.set_xlabel(r’$\Phi$’)
ax2.set_ylabel(r’$\eta$’)
ax2.legend(loc=’best’)

plt.tight_layout()

• The log-log plot is the more useful way to see the behavior.

• For Φ  1 the effectiveness factor is practically one. That means the

diffusion rate is much faster than reaction, so the concentration in the
pellet is practically uniform and equal to the surface concentration.

• For Φ  1 then η ≈ 1/Φ. Near Φ = 1 you may want to evaluate the

actual solution.

2 The effectiveness factor in action

We start by computing the effectiveness factor for the particles.

• Φ = k(R/3)2 /DA
p

• η = Φ1 tanh1 3Φ − 3Φ
1
 

Rp = 0.3 # particle radius cm

k = 2.6 # 1/s
De = 0.007 # cm^2/s

a = Rp / 3
Phi = np.sqrt(k * a**2 / De)
eta = 1 / Phi * (1 / np.tanh(3 * Phi) - 1 / (3 * Phi))
print(r’\eta = {}’.format(eta))

9
 Next we setup the mole balance. We have a simple problem:
dF a
dV = rA = ηkCA
We can express this as an integral:
R 0.03F
V = F a0 A dF rA
a

An important question though is what volume are we talking about here?

It is the volume inside the pellets. We can estimate the mass of pellets
required by multiplying this volume by the pellet density. Then we can
divide that mass by the bed density to
Vreactor = Vparticles ∗ ρp /ρb

T = 450
R = 82.06 # Gas constant cm^3 atm / mol / K
Pa = 1.5 # atm

Ca0 = Pa / R / T # mol / cm^3

Fa0 = 12 # mol / s

v0 = Fa0 / Ca0

rho_b = 0.6 # bed density g / cm^3

rho_p = 0.85 # particle density g / cm^3

from scipy.integrate import quad

def integrand(Fa):
Ca = Fa / v0
r = eta * k * Ca
ra = -r
return 1 / ra

V_pellets, _ = quad(integrand, Fa0, 0.03 * Fa0)

M_pellets = V_pellets * rho_p
V_bed = M_pellets / rho_b
print(’The bed volume is {:1.2e} cm^{{3}}’.format(V_bed))

3 Summary
• Mass transfer reduces the effectiveness of a catalyst by reducing the
concentration of reactants in the pores

10
• We estimate the effective rate by integrating over the volume

• This allows us to use mole balance framework to still do reactor design

• We can quickly estimate the effectiveness factor by knowing the Thiele

modulus (Φ)

11