Session 1613

Reactor Design With Matlab In a Manufacturing Environment

Dr. Charles U. Okonkwo

Arizona State University

ABSTRACT

The motivation for this study arises from a class project in an Alternative Energy course
MET 494. A professor with mechanical processing background taught the course to students with
similar background during the 1996 fall semester. During the 1996 spring semester, the
professor’s MET 494 students produced hydrogen in a batch reactor via a methane steam
reforming reaction on a nickel catalyst. The batch reactor temperature was about 800 0C and
pressures varied between 85 and 97 psig. The class objective, among other things, was to
produce hydrogen in a continuous flow reactor and understand the behavior of such a reactor.

Hydrogen is a promising fuel alternative. As an additive, hydrogen may boost the

performance of jet propulsion engines. Some auto manufacturers have begun research in this
area. For example, Mazda has already produced a hydrogen fueled prototype engine that
outperformed electric prototype engines. Diamler-Benz is also researching the use of hydrogen
in internal combustion engines. Additionally, the students’ future objective is to research the use
of hydrogen as a stand alone alternative in jet propulsion and internal combustion engines. The
professor asked me to help design a reactor, which the students would build.

I have modeled hydrogen production using a packed bed reactor. The design equations
consist of coupled material and energy balances. Rate kinetics used in the design equations were
obtained from the literature. The design equations contain one endothermic reaction and one
exothermic reaction, yielding an overall reaction that is endothermic. I used Matlab to solve the
resulting five nonlinear ordinary differential equations and an algebraic equation. Using the
model the students can simulate production of hydrogen by adjusting reactor length, area, heating
of reactants, molar flow rates of methane and steam, inlet temperature, inlet pressure and obtain
hydrogen yield and flow rates of by-products. It uses a non temperature dependent rate constant
obtained from the literature, and can be modified to handle a temperature dependent rate
constant. The model gives a decreasing temperature profile across reactor length. The model can
handle pressure calculations by simply adding an extra differential equation involving pressure as
a dependent variable. However, it assumes constant pressure due to cost considerations. The
model empowers the student to build and study the methane steam reforming reactor and gain
better insight.
Page 2.340.1

1
Introduction

The promise of hydrogen as a fuel for automobile and jet propulsion engine has sparked
interest in hydrogen production. This opinion is shared by Marr (1). Steam methane
reforming (SMR) is the method of hydrogen production described in this study.
According to Rosen and Scott (2), it is one of the most important industrial processes for
hydrogen production today. Rosen and Scott (1) describe the status of SMR process to be
a mature technology. Though the process involves both exothermic and endothermic
reactions, the net reaction is endothermic. The energy required to promote the reaction is
supplied by heat from the exhaust of an automobile engine and a built-in heater inside the
reactor that can be turned off and on. The MET 494 students have background in metal
working based manufacturing. I undertook the modeling to enhance the students’
knowledge regarding the behavior of the reactor, at least in a qualitative manner. The
model allows the simulation of the reactor via parameters such as cross-sectional area of
reactor, molar flow rates of reaction components, built-in heating, q inside the reactor,
inlet temperature of reactants and reactor length. Due to unforeseen circumstances, the
students completed building the reactor at the end of the semester and had no time to run
the experiment. I have simulated hydrogen production on the computer using matlab.
This should prove to be a valuable tool in running the hydrogen production experiment.

Reactions Within the Packed Bed Reactor/Theory Behind the Packed bed Reactor

The two reactions involved in the reforming of natural gas over a nickel catalyst are given
below:

CH4 + H2O → CO + 3H2 (1)

∆Hrxn = +206 kJ mol-1

CO + H2O ⇔ CO2 + H2 (2)

∆Hrxn = -41 kJ mol-1

The reforming reaction, equation (1) is highly endothermic, while the water gas shift
reaction, equation (2) is exothermic. The overall reaction is endothermic. The reforming
reaction is far from equilibrium, whereas the shift reaction is very close to equilibrium.
Agnelli et al. (3).

Theory Behind the Packed bed Reactor

Modeling Equations for Packed Bed Reactor

Table 1 , a listing of reactor variables and their respective units, facilitates readability of
the model equations.
Page 2.340.2

2
Table 1

Variable Name Units

A Methane none

B Water none

C Carbon Monoxide none

D Hydrogen none

E Carbon Dioxide none

T Temperature kelvin

S Cross sectional area of reactor (meter)2

Cpi Specific heat of component i joule/(g mole • kelvin)

rA Reaction rate expression g mole/ (volume of catalyst • minute)

ρb Catalyst bulk density gram catalyst/ volume of catalyst

q Heating inside reactor joule/(minute • meter)

k Reaction rate constant g mole/(gram catalyst • minute • atmosphere)

Pi Partial pressure of component i atmospheres

PT Total pressure atmospheres

ni Molar flow rate component i g mole/ min

nT Total molar flow rate g mole/ min

K(T) Equilibrium constant none

KA Constant taken from Agnelli’s data none

KB Constant taken from Agnelli’s data (atmospheres)-1

z Reactor length meters

Page 2.340.3

3
Assuming steady state, the requisite equations for the model are given below:

dn A
= − S rA , n A (0) = n A 0 (3)
dz

dnB
= − S rA , nB (0) = nB 0 (4)
dz

dnC
= S rA , nC (0) = nC 0 (5)
dz

dnD
= 3S rA , nD (0) = nD 0 (6)
dz

dT − S ( ∆Hrxn )rA q
= + , T(z=0) = T0
dz n A cPA + nB cPB + nC cPC + nD cPD n AcPA + nB cPB + nC cPC + nD cPD
(7)

nE nD
K (T ) = (8)
nC nB

where the reaction rate, rA is given by:

kPA ρb n P
rA = , where Pi = i T , and subscript " i " = A, B , C or D (9)
P nT
(1 + K A B + K B PC ) n
PD
n
kPT A ρb
nT
rA = , where PT is total pressure (10)
nB nC n
(1+ K A + KB PT )
nD nT

In general, k is a function of T. Agnelli et al. (3) lists three values of k, KA, KB at

different temperatures for different reaction rate models, and different power, n values.
This suggests that a regression technique can be used to write k, KA, KB, each as a
function of T. Since we are more interested in the qualitative behavior of the reactor, a
first level approximation using constant values of k, KA, KB is sufficient. Agnelli et al. (3)
indicates that the above reaction rate expression with power, n=1 is adequate for
describing hydrogen production via methane steam reforming. Matlab is the application
package used in solving the above model equations 3 through 10 for different
combinations of reactor parameters. I have shown below an example Matlab program for
Page 2.340.4

certain combinations of reactor parameters.

4
Matlab Program

global PT S DELH Q CPA CPB CPC CPD NA NC NB ND NE NT

PT = 1.5; % PT is total pressure in atmospheres
S = 0.05; % S= 0.5 square meters of area
DELH = 206014; % DELH is heat of reaction in Joules per gram mole
Q = 5000; % Heat input rate per unit lenth of bed (Joules/min-meter)
CPA = 36.9607; % specific heat (J/gmole-K)
CPB = 33.7295; % specific heat (J/gmole-K)
CPC = 29.1668; % specific heat (J/gmole-K)
CPD = 28.6455; % specific heat (J/gmole-K)
zz0 = 0.0;
zzf = 0.5;
YY0 = [3;3.5;0;0.000001;1300];
[zz,YY] = ode23('chrlspr3',zz0,zzf,YY0);
K = 0.45; % For now, K is assumed to be independent of temperature, K = constant
NA= YY(:,1);
NB= YY(:,2);
NC= YY(:,3);
ND= YY(:,4);
T = YY(:,5);
NE= .45* YY(:,3).* YY(:,2)./YY(:,4);
NT= NA+NB+NC+ND+NE;
plot(zz,NA,'r+',zz,NB,'g-')
title('MOLES/MIN versus REACTOR LENGTH')
xlabel(' Reactor Length')
ylabel('Gram-moles/min')
grid
pause, close
plot(zz,NC,'b-.',zz,ND,'y--',zz,NE,'r+')
title('MOLES/MIN versus REACTOR LENGTH')
xlabel(' Reactor Length')
ylabel('Gram-moles/min')
grid
pause, close
plot(zz,T)
title('TEMPERATURE versus REACTOR LENGTH')
xlabel('Reactor Length')
ylabel('TEMPERATUTE 0K')
grid
Page 2.340.5

function W =chrlspr3(zz,YY)

5
global PT S DELH Q CPA CPB CPC CPD NA NB NC ND NE NT T
S= 0.01
NT =YY(1)+YY(2)+YY(3)+YY(4)+(0.45* YY(3)* YY(2)/ YY(4));
NTCP=YY(1)*CPA + YY(2)*CPB + YY(3)*CPC + YY(4)*CPD;
W(1)=-1.02*(10^6)*S*(0.10949*PT*YY(1)/NT)/(1+0.239*YY(2)/YY(4)
+17.62*YY(3)*PT/NT);
W(2)=-1.02*(10^6)*S*(0.10949*PT*YY(1)/NT)/(1+0.239*YY(2)/YY(4)
+17.62*YY(3)*PT/NT);
W(3)= 1.02*(10^6)*S*(0.10949*PT*YY(1)/NT)/(1+0.239*YY(2)/YY(4)
+17.62*YY(3)*PT/NT);
W(4)=3*1.02*(10^6)*S*(0.10949*PT*YY(1)/NT)/(1+0.239*YY(2)/YY(4)
+17.62*YY(3)*PT/NT);
W(5)=(-S*DELH*W(3)+Q)/(NTCP);

Results and Discussion

A constant value of 0.45 has been used for the equilibrium constant K(T). A low value of
K is required, because the shift reaction, equation (2) is an equilibrium reaction. Twigg
(4) lists equilibrium constants at various temperatures in appendix 7 of his book. Also, a
constant value of ∆H rxn is used in this first level approximation. The bulk density of
nickel catalyst , ρ b , used in this simulation is 1.02*106 gram catalyst/(meter3). In order to
prevent division by zero during the first iteration steps in Matlab, the initial condition on
hydrogen molar flow rate, nD0 is set to 0.000001. An example set of parameters is shown
below:
PT = 1 atmosphere
S ∆H rxn ratio q k K KA KBz T0
0.01 206014 3.5/3 5,000 0.10949 0.45 0.239 17.62 0.5 1300
0.01 206014 13.5/3 5,000 0.10949 0.45 0.239 17.62 0.5 1300
0.01 206014 3.5/3 50,000 0.10949 0.45 0.239 17.62 0.5 1300

Figures 1 through 3 depict the graphs resulting from the simulation of the first
combination of parameters, figures 4 and 5 show those from the second combination of
parameters, and figure 6 show those resulting from the third combination of parameters.
It is obvious that the number of possible combinations is very large. Therefore, in order to
understand the behavior of the reactor, it suffices to show graphs of a few of the
simulations.

All plots show that a reactor with a cross-sectional area of 0.01 (meters)2 and a length
between 0.25 and 0.3 meters will yield almost complete conversion for the given feed
(water/methane) ratio. Figures 3 and 5 indicate that for the case of moderate heating,
q=5,000 joules/(meter • minute), the reactor temperature does not rise above the inlet feed
temperature. This suggests that most of the heat from the heater is used in promoting the
endothermic reaction. For very high heating, q=50,000 joules/(meter• minute), figure 6
Page 2.340.6

indicates that the temperature is rising. For much larger q values, not displayed, the

6
reactor temperature actually rises above the feed temperature without any appreciable
change in conversion. This suggests that excess heat is used to raise the reactor
temperature. Figures 2 and 4 indicate that a high molar ratio (13.5/3) of water to methane
promotes the formation of carbon dioxide in the shift reaction. Therefore, the rate of
hydrogen formation is slower for the case of high molar ratio. This suggests that an
optimal ratio, which gives the best performance, exists. In practice, we would find this
ratio experimentally.

Conclusion
The model is capable of simulating various combinations of reactor parameters. As a first
approximation model, it is useful for designing a steam methane reactor for the
production of hydrogen. It provides reactor size for certain feed rates. A more accurate
model may be obtained by inputting k, K, ∆H rxn as functions of temperature into the
Matlab program.

References
(1) Marr, A. 1996. Hydrogen Powered Rotaries. Website. http:www.monito.com/wankel/hydrogen.html.

(2) Rosen, M.A. and Scott D.S., 1986. Analysis and Comparison of the Thermodynamic Performance of
Selected Hydrogen Production Processes. Can. Proc. Intersoc. Energy Convers. Eng. Conf., 21st, vol. 1,
pp. 266-271.

(3) Agnelli, M.E., Ponzi, E.N., and Yeramian, A.A., 1987. Catalytic Deactivation on Methane Steam
Reforming Catalysts. 2. Kinetic Study. Ind Eng. Chem. Res. vol 26, pp. 1707-1713.

(4) Twigg, M.V. Catalyst Handbook pp.225-238. Wolfe Publishing Ltd., England 1989.

Page 2.340.7

7
 GRAM MOLES/MIN versus REACTOR LENGTH
3.5

2.5
Gram-
2
moles/
min
1.5

1
B

0.5
A
0
0 0.1 0.2 0.3 0.4 0.5
Reactor Length meters
Figure 1: Ratio 3.5/3, q=5000, S =.01

GRAM MOLES/MIN versus REACTOR LENGTH

9
D
8
7
Gram- 6
moles/
min 5
4
C
3
2
1
E
0
0 0.1 0.2 0.3 0.4 0.5
Reactor Length meters
Figure 2. Ratio: 3.5/3, q=5,000, S =.01
Page 2.340.8

8
 TEMPERATURE versus REACTOR LENGTH
1300

1280
T
e
m 1260
p
e 1240
r
a
t 1220
u
r 1200
e
K
1180
0 0.1 0.2 0.3 0.4 0.5
Reactor Length meters
Figure 3. Ratio: 3.5/3, q=5,000, S=.05

GRAM MOLES/MIN versus REACTOR LENGTH

10

8
D
Gram-
moles/
min 6

4
C

2
E

0
0 0.1 0.2 0.3 0.4 0.5
Reactor Length meters
Figure 4. Ratio 13.5/3, q=5000, S=0.05
Page 2.340.9

9
 TEMPERATURE versus REACTOR LENGTH
1310

T 1300
e
m 1290
p
e
r 1280
a
t 1270
u
r
e 1260
K
1250
0 0.1 0.2 0.3 0.4 0.5
Reactor Length meters
Figure 5. Ratio: 13.5/3, q=5,000, S=.05

TEMPERATURE versus REACTOR LENGTH

1360

1350
T
e 1340
m
p 1330
e
r
1320
a
t
u 1310
r
e 1300
K
1290
0 0.1 0.2 0.3 0.4 0.5
Reactor Length meters
Figure 6. Ratio: 3.5/3, q=50,000, S =0.01
Page 2.340.10

Charles U. Okonkwo

10
 Dr. Charles U. Okonkwo graduated with bachelors and masters degrees in chemical engineering from
Iowa State University, and a Ph.D. in chemical engineering from the University of Florida. He has worked
as a process engineer for both the chemical and semiconductor industries. Since joining the College of
Technology and Applied Sciences at Arizona State University as a lecturer, he has taught graduate courses
in hazardous waste management and undergraduate courses in the Department of Manufacturing. Prior to
joining the College of Technology and Applied Sciences, he taught for several years in the Department of
Mathematics at Arizona State University.

Page 2.340.11

11