Fixed-Bed Reactor Modeling and Simulation with eLearning Tools

Margarida J. Quina 1, A. Cristina Almeida-Costa 2, Rosa M. Quinta-Ferreira 1

Abstract - Fixed-bed reactors (FBR) are largely used in
industrial catalytic processes requiring a careful
understanding at the undergraduate level of Chemical
Engineering. Several mathematical models are available
to represent the behavior of such heterogeneous systems
with various degrees of complexity, including different
transfer mechanisms at the fluid level or inside the
catalyst particles. This will permit the application of the
fundamental concepts of the transport mechanisms
previously learned by the students in the first years of the
course. One-dimension and two dimension models with
or without dispersion effects in the axial and radial
directions may lead to transient and steady state models,
while the heterogeneity of the system can be simplified by
assuming a pseudo-homogeneous behavior, which
neglects inter and intraparticle resistances. Regarding
the most complete heterogeneous models accounting for
external and internal mass and thermal gradients, the
behavior of the catalyst particles is fundamental to follow
the multiscale characteristics of such systems. This, is in
agreement with the New Frontiers in Chemical
Engineering Education that claim systems teaching
embracing the understanding of the physical and
chemical phenomena in a broad scale. In this context,
several intraparticle models including diffusional and
convective transports can be developed and integrated in
the global model of the fixed bed reactor. The different
mathematical models are represented by partial or
ordinary differential equations associated to algebraic
expressions, requiring numerical solutions through
various computational codes. Therefore, the great
interest associated to the teaching of these industrial
systems is frequently inhibited by their complexity, so
that modern tools based on e-learning methodologies will
be much welcome. In this context, the global objective of
the present work addresses the construction of a platform
for visualizing the general behavior of fixed-beds
involving reactants and products profiles and thermal
gradients as well.
Index Terms - Chemical Processes, E-learning, Fixed-bed
reactors, Modeling.
INTRODUCTION
The industrial chemical processes convert raw materials in
products, and the reactor is often considered as the heart of
the plant. The selection and design of the reaction units are
essential for the economic success of a chemical industry,
imposing final yields and conversions. The Chemical
1
2

Reaction Engineering is then an area of Chemical

Engineering with a significant role, and our main goal focus
on the development of e-learning tools in the domain of
heterogeneous processes, specifically fixed-bed reactors
(FBR). This work addresses catalytic reactional systems with
particular emphasis in fixed-bed reactors that are largely
used in industry. These processes may be studied through
diverse mathematical models: pseudo-homogeneous and
heterogeneous models in steady state or transient regimes [13]. For real systems, in general, no analytical solutions are
available, and different numerical techniques are required for
computational simulation. The ordinary differential
equations, ODEs, and/or partial differential equations, PDES,
may be solved by using various commercial Fortran codes:
GEAR, PDECOL, FORSIM, DDASAC, DASSL.
The analysis of the behavior of FBR is proposed in the elearning platform in two levels:
- general analysis through different mathematic models,
including the catalyst particle and the film resistances;
- two case studies involving the synthesis of phthalic
anhydride and formaldehyde through partial oxidation of
ortho-xylene and methanol, respectively.
In particular, the operation of these systems highlights the
thermal instabilities that are possible to occur in such
exothermic processes, namely temperature runaway and
wrong way behavior, relevant for safety. The virtual platform
allows the students to understand the effect of different
operating conditions on the FBR behavior. Also, some
complex theoretical concepts can be easily understood since
the time consuming for mathematical calculations are almost
instantaneous. The e-learning philosophy has then important
advantages when compared with classical teaching, since is
embedded into a web page permitting a broad dissemination
of the behavior of industrial systems, often impossible to
teach in conventional classes.
MATHEMATICAL MODELS
The behavior of fixed-bed reactors (FBR) may be
studied by using several mathematical models, depending on
the complexity that is taken into account for, and Figure 1
resumes the different possibilities. For example, PH1D
model correspond to a pseudo-homogeneous approach in one
dimension, and in this case the film and the particle are
implicitly considered. In the case of heterogeneous models,
the gradients in the film and in the catalyst may be taken into
account (HT1Dd, HT1Ddc, HT2Dd, HT2Ddc). The model
HT2Dd in transient conditions includes equations (1) to (12),
indicated in Table 1, and corresponds to a complex approach
for simulating the dynamic behavior of FBR.

Chemical Engineering Department - University of Coimbra, [email protected]; [email protected];

Companhia Industrial e Resinas Sintticas, Cires S.A, [email protected]

Coimbra, Portugal

September 3 7, 2007
International Conference on Engineering Education ICEE 2007

TABLE 1
TWO DIMENSION DYNAMIC HETEROGENEOUS MODEL WITH DIFFUSIONAL TRANSPORT IN THE CATALYST (HT2DD)

Fluid phase in the reactor

Mass balance to specie i :

Energy balance: f Cp f

C i ,b
t

= u i

C i ,b
z

+ Dea

2 C i ,b 1 C i ,b
+ D er
+
r 2
r r

2 C i ,b
z

+ r
i i ,b

(1)

2Tb 1 Tb
Tb
Tb
2Tb

+ (H j ) j r j ,b
= u i f Cp f
+ ea
+

+
er
2
t
z
r r j
z 2
r
z ,r ; t = 0

Initial Conditions:

(2)
(3)

C i ,b = C i0,b ; Tb = Tb0

Boundary Conditions:

u i (C i ,0 C i ,b ) = Dea

t , r ; z = 0
t , r ; z = L

t , z ; r = 0

t , z ; R0 = 0

C i ,b
z=L

C i ,b
r

=0 ;
r =0

C i ,b
r

Tb
z

=0 ;

C i ,b

Tb
r

u i f Cp f (T0 Tb ) = ea

;
z =0

Tb
z

z =0

(4)

=0
z =L

=0
r =0

= 0 ; er
r = R0

Tb
r

(5)

= hw (Tb T w )
r = R0

Fluid/solid interface

k f ,i a p (C i , s C i ,b ) = ri ,obs

(6)

h f a p (Ts Tb ) = ( H j ) r j ,obs

(7)

Mass balance to specie i:

Energy balance:

Catalyst particle

Mass balance to specie i:

C i , p
t

2 C i, p
s C i , p
= D e ,i
+
r 2
r p r p
p

+ r
i, j j , p

j

(8)

2T p
s T p
= e
+
+ ( H j ) r j , p
r 2 r p r p
t
p
j

T p

Energy balance:

( Cp ) p

Initial Conditions:

rp ; z ; t = 0

Boundary conditions:

t , z; r p = R p

(10)

C i , p = C i0, p ; T p = T p0
C i, p = C i,s ; T p = Ts
C i , p

t , z; r p = 0

(9)

r p

=0 ;
rp = 0

T p
r p

=0

(11)

rp = 0

Effectiveness factor referred to bulk conditions

robs
=
=
r (C b , Tb )

1
Vp

r (C p , T p ) dV p
(12)

r (C b , Tb )

Coimbra, Portugal

September 3 7, 2007
International Conference on Engineering Education ICEE 2007

One- dimension model

- with/without
axial dispersion
- with/without film
gradients
- with/without
particle gradients

Two- dimensions model

PH1D

HT1Dd/
HT1Ddc

PH2D

HT2Dd/
HT2Ddc

not
not
applicable
applicable




not
not
applicable
applicable
FIGURE 1
STEADY-STATE MODELS FOR FIXED BED REACTOR.

This general model is presented to students in

classroom, and the meaning of each term is fully explained.
Starting from this model (HT2D) the simplest ones are easily
obtained. The meaning of the different symbols is indicated
in the notation section, at the end of the paper.
The subscripts coupled to HT1D and HT2D are referred
to the mass mechanism inside the catalyst (Figure 1).
Therefore, when only the diffusion flow is considered the
reference is HT1Dd or HT2Dd. In the case of both
intraparticle convection and diffusion flows are considered,
the models are referred HT1Ddc or HT2Ddc. The axial
dispersion can be included in any case (PH1D, HT1Dd/
HT1Ddc, PH2D and HT2Dd/ HT2Ddc).

SIMULATION METHODS

following scheme of consecutive/parallel pathways is used to

describe this chemical process, where A stands for o-xylene,
B for phthalic anhydride and C for the final oxidation
products (CO and CO2) [12]:
A

(+ air), K1
(+ air),K3

(+ air), K2

Due to the large excess of oxygen, the rate equation for each
step j (j=1,2,3) is considered to be of pseudo-first-order:
rj = K j pi pO2 , where i is the corresponding reactant (i=A,B).
Two one-dimensional heterogeneous models are used for the
description of the system, one that accounts for both
intraparticle mechanisms of convection and diffusion
(HT1Ddc) inside the catalyst pellets and the other considering
only the diffusive flow (HT1Dd). Axial dispersion of mass
and heat was neglected and a constant wall temperature
(equal to the inlet fluid temperature) was assumed in the
fixed-bed. An isothermal catalyst with slab geometry was
also considered. The steady-state equations and the reactor
and kinetic parameters may be found elsewhere [7]. The
resulting system of differential-algebraic equations is
integrated by using the commercial software code DDASAC
- Double Precision Differential Algebraic Sensitivity Code
[13].
From the resolution of the equations, several results may be
simulated, for the analysis of different perspectives of the
reactor steady-state behavior. For example, Figure 2(a)

The e-learning platform for teaching FBR will be

included in a most global one referred Virtual Laboratories
of Chemical Processes, and in particular, in a sub-level
Chemical Reaction. The virtual platform will be supported
by interactive simulation modules, and the web interface
allows the students entering data for simulations. The outputs
are mainly graphics or numerical results.
The Matlab and Fortran are used as computational
platforms for all simulation concerning FBR. Depending on
the type of differential equations referred above the models
may be solved using commercial Fortran codes such as
GEAR, PDCOL, FORSIM, DDASAC, or routines from the
Matlab.
As referred in the introduction, the students may try
simulations in the e-learning platform by two main
approaches: a general analysis through different mathematic
models, including the catalyst particle and the film
resistances and based on two case studies of industrial
interest.
CASE STUDIES
Two case studies are available for students, aiming the
understanding of real industrial processes: partial oxidation
of ortho-xylene to phthalic anhydride [4-7] and partial
oxidation of methanol to formaldehyde [8-11].
I. Case Study 1 Synthesis of Phthalic Anhydride
This case study refers to the partial oxidation of o-xylene to
phthalic anhydride in a fixed-bed reactor packed with a
V2O5-TiO2 catalyst supported on large-pore particles. The

FIGURE 2
EVOLUTION OF FLUID TEMPERATURE AND CONCENTRATION OF
O-XYLENE (A) AND PHTHALIC ANHYDRIDE (B) ALONG THE
REACTOR AXIAL COORDINATE.

Coimbra, Portugal

September 3 7, 2007
International Conference on Engineering Education ICEE 2007

shows the bulk temperature profiles and Figure 2(b) the oxylene and phthalic anhydride dimensionless concentration
profiles along the reactor axial coordinate for different values
of the intraparticle mass Peclet number, m, at the inlet
conditions T0=625K and P0=0,01 atm. The intraparticle and
interfacial profiles of the dimensionless concentration of oxylene and phthalic anhydride calculated at the reactor inlet
(z*=0) are also depicted in Figures 3.

the evolution of the differential yield, , and the differential

selectivity of phthalic anhydride, S, along the reactor axial
coordinate, calculated at the same conditions of Figure 2.
The results indicated in Figures 2 to 4 are examples, which
students may simulate interactively, being possible to change
the operating conditions of the system. The effect of those
changes is observable in graphical representations.
II. Case Study 2 Synthesis of Formaldehyde

FIGURE 3
INTRAPARTICULAR
AND
INTERFACIAL
DIMENSIONLESS
CONCENTRATION PROFILES OF O-XYLENE (A) AND PHTHALIC
ANHYDRIDE (B) AT THE REACTOR ENTRANCE (Z*=0) FOR
DIFFERENT VALUES OF M (T0=TW=625K AND P0=0,01 ATM).

For chemical processes involving competitive reactions,

yield and selectivity are very important parameters for the
evaluation of the reactor efficiency. Figures 4 (a)-(b) show

The system analyzed under this case study was the partial
oxidation of the methanol to formaldehyde over
iron/molybdenum oxides catalyst. The kinetic model
involves two reactions: the partial oxidation of methanol to
formaldehyde, which may be followed by a side reaction
involving the partial oxidation of the formaldehyde to CO
and H2O. The kinetic model may be found in [10] and the
operating conditions used by default correspond to the
industrial ones (1.4 atm; 257C; 7.5% for feed methanol
mole fraction). Although the industrial production of
formaldehyde involve FBR with thousands of tubes, in the
simulations only a single tube is considered with a length of
0.75 m and divided into two zones: 0.2 m of diluted catalyst
with 50% inert in the first region (less active bed) and 0.55 m
of pure catalyst. This distribution of the catalyst allows mild
heat generation, and then lower hot spots are obtained. The
e-learning platform allows the observation of:
- steady state profiles for concentrations and temperature,
according to different mathematical models (PH1D,
HT1D and HT2D). For heterogeneous models, the mass
flux inside the particle may account for only diffusion
(HT1Dd; HT2Dd) or both diffusion and convection
(HT1Ddc; HT2Ddc).
- transient behavior of the reactor according PH1D, HT1Dd
and HT1Ddc.
Figure 5 and 6 show as example, some axial profiles of
temperature in the case of steady state conditions and for
dynamic conditions, respectively.

670

HT2Ddc(radial mean)

650

Ts(K)

630

PH1D

HT1Ddc

610

HT2Dd(radial mean)

590

HT1Dd

570
550
530

0.1

0.2

0.3

0.4
z(m)

0.5

0.6

0.7

FIGURE 5
AXIAL SOLID TEMPERATURE PROFILES PREDICTED WITH
DIFFERENT MATHEMATICAL MODELS: PH1D, HT1DD, HT1DDC,
HT2DD, HT2DDC

FIGURE 4
DIFFERENTIAL YIELD, , AND THE DIFFERENTIAL SELECTIVITY
OF PHTHALIC ANHYDRIDE, S, ALONG THE REACTOR AXIAL
COORDINATE.

Coimbra, Portugal

September 3 7, 2007
International Conference on Engineering Education ICEE 2007

CONCLUSIONS

820
770
720
T(K)

Initial
Final steady-state

670
620
570
520

0.1

0.2

z*

0.3

0.4

0.5

(a)
650

E-learning tools may be the key for the success of

students to understand the behavior of FBR. In fact, due to
the high complexity of the mathematical equations involved
in the description of these systems, the classical approach for
teaching this engineering subject sometimes fails, since only
very simplified case studies may be solved in the classroom.
Through simple computational simulations, made in a
friendly web-based interface, the performance of a FBR is
easily obtained for a given set of operation conditions.
Moreover, the effect of changes in the operating conditions
may also be straightforwardly understood, based on simple
graphical representations of the system behavior.
To conclude, it is important to note that e-learning tools
may lead to a reduction of the number of classical lectures,
being students more independent, which is in accordance
with the methodologies stated by the Bologna Process.

630

Initial

ACKNOWLEDGMENT

Final steady-state

T(K)

610

The authors acknowledge the financial support from POSC

(Programa Operacional da Sociedade de Conhecimento,
Portugal), contract 743/4.2/C/REG.

590
570
550
530

NOTATION
0

0.1

0.2

(b)

0.3

0.4
z*

0.5

0.6

0.7

0.8

FIGURE 6
TRANSIENT AXIAL TEMPERATURE PROFILES OBTAINED
WITH PH1DT MODEL FOR A FIXED BED PACKED WITH (a)
PURE CATALYST, (b) DILUTED CATALYST, BY DECREASING
THE FEED TEMPERATURE -5% (TO=557530 K).

These mathematical models are based on common

assumptions: constant pressure in the reactor, constant values
for physical and transport properties, constant wall
temperature, constant bed porosity, flat radial velocity
profiles (plug flow) and constant catalyst activity in each
catalytic zone. The simulated mathematical equations may be
found in [8] for the case of steady state and in [10] for
transient conditions. The commercial code DDASAC
package was used for solving the steady state [13-14], which
is an extension of DASSL code, developed by Petzold [15].
The transient model was solved with the PDECOL [16]. The
numerical solutions involve complex strategies which were
discussed in detail elsewhere [11].
It is important to note that the dilution of the catalyst bed
in the first part of the tubes has advantages concerning the
parametric sensitivity and runaway behavior [9].
In this case study students may test the influence of
several variables such as the operating conditions
(concentration, temperature), the wall temperature, the inlet
pressure, etc. In transient conditions, different changes may
be virtually experimented (steps, ramps, etc.). Moreover, the
influence of the dilution of the catalyst can be easily
understood, mainly in thermal profiles, by simulating this
important industrial process.

ap - specific particle area (m-1)

Ci - concentration of species i (mol/ m3)
C0 - o-xylene feed concentration, (mol/m3)
Cp - heat capacity (J/kg K)
De - effective diffusivity (m2/s)
dp - particle diameter (m)
fi - dimensionless concentration of species i , Ci/C0
hf - film heat transfer coefficient (J/ m2 s K)
hw - wall heat transfer coefficient (J/ m2 s K)
kfi - film mass transfer coefficient of species i (m/s)
L - reactor length (m)
r - reactor coordinate (m)
rj - rate of reaction j (mol/ s kgcat)
robs,i - observed reaction rate inside the catalyst (mol/s kgcat)
rp - particle coordinate (m)
Rp - half thickness of the catalyst particle (m)
R0 - tube radius of the reactor (m)
s - catalyst shape factor (slab: s=0; cylinder: s=1; sphere:s=2)
t*- dimensioless time (t/)
t - time (s)
T- absolute temperature (K)
T0 - feed temperature (K)
ui - interticial velocity (m/s)
v0 - intraparticle fluid velocity (m/s)
z - reactor axial coordinate (m)
z*- dimensioless reactor axial coordinate (z/L)
Greek symbols
i,j - stoichiometric coefficient of specie i, in the reaction j
Hj - enthalpy of reaction j (J/mol)
- bed porosity
e - effective thermal conductivity (J/m s K)

Coimbra, Portugal

September 3 7, 2007
International Conference on Engineering Education ICEE 2007

j - effectiveness factor of reaction j

- density (kg/m3)
space time for the fluid (L/ui)
Subscripts
a- axial
b - bulk
f - fluid
i- chemical specie
j - reaction
p - particle
r - radial
s - particle surface
w - wall
REFERENCES
[1] G. F. Froment, K.B. Bischoff, Chemical reactor analysis and design,
Wiley, 2nd ed., New York, 1990, pp. 664.
[2] S. Feyo de Azevedo, M. Romero-Ogawa, A.P. Wardle, Modelling of
tubular fixed-bed catalytic reactors: a brief review, Trans. IChemE,
68A, pp. 483-502, 1990.
[3] S. Pereira Duarte, G.F. Barreto, N. Lemcoff, Comparison of
twodimensional models for fixed bed catalytic reactors, Chem. Eng.
Sci., 39, pp. 1017-1024, 1984.
[4] R.M. Quinta-Ferreira, A.C.A. Costa, A.E. Rodrigues, Dynamic
Behavior of Fixed-Bed Reactors with Large-Pore Catalysts Bidimensional Heterogeneous Diffusion/Convection Model, Comp.
Chem. Eng., 16, 8, pp. 721-751, 1992.
[5] C.A. Almeida-Costa, R.M. Quinta Ferreira, A.E. Rodrigues, WrongWay Behavior in Fixed-Bed Reactors with Large-Pore Catalysts,
Chem. Eng. Sci., 49, pp. 5571-5583, 1994.
[6] R.M. Quinta Ferreira, C.A. Almeida Costa, A.E. Rodrigues, Effect of
Intraparticle convection on the Transient Behaviour of Fixed-Bed
Reactors: Finite Differences and Collocation Methods for Solving
Unidimensional Models, Comp. Chem. Eng., 20, pp. 1201-1225, 1996.
[7] Almeida-Costa, A.C.A. Estudo do comportamento estacionrio e
dinmico de reactores catalticos de leito fixo com conveco
intraparticular, PhD Thesis, University of Coimbra, 2005.
[8] M.J. Quina, R.M. Quinta Ferreira, Model Comparison and
Sensitivity Analysis for a Fixed Bed Reactor with Two Catalytic Zones,
Chem.Eng.J., 75, 3, pp. 149-159, 1999.
[9] M.J. Quina, R.M. Quinta Ferreira, Thermal Runaway Conditions of
a Partially Diluted Catalytic Reactor, Ind.Eng.Chem.Res., 38, 12,
pp. 4615-4623, 1999.
[10] M.J. Quina, R.M. Quinta Ferreira, Start-Up and Wrong-Way
Behaviour in a Tubular Reactor: Dilution Effect of the Catalytic Bed,
Chem.Eng.Sci., 55, 18, pp. 3885-3897, 2000.
[11] M. J. Quina, R.M. Quinta Ferreira, Simulation of a Fixed Bed
Reactor with Two Catalytic Zones: Numerical Methods, in Proc. 2 nd
Meeting on Numerical Methods for Differential Equations,
Coimbra, 1998, pp. 447-456.
[12] G.F. Froment, Fixed bed catalytic reactors: current design status. Ind.
Eng. Chem., 59, 18-27, 1967.
[13] W.E. Stewart, M. Caracotsios, J.P. Srensen, Software package
documentation of DDASAC - Double Precision Differential Algebraic
Sensitivity Code. University of Wisconsin-Madison, EUA., 1991.
[14] M. Caracotsios, W.E. Stewart, Sensitivity analysis of initial value
problems with ODEs and algebraic equations, Comp. Chem. Eng. , 94,
pp. 359-365, 1985.
[15] L.R. Petzold, Differential/algebraic equations are not ODEs, Siam J.
Sci. Stat. Comput. , 3, pp. 367-385, 1982.
[16] N. Madsen, R. Sincovec, PDECOL: General collocation on finite
elements, Chem. Eng. Sci, 30, pp. 587-596, 1975.

Coimbra, Portugal

September 3 7, 2007
International Conference on Engineering Education ICEE 2007