Download as pdf or txt
Download as pdf or txt
You are on page 1of 28

Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

https://doi.org/10.1007/s11144-019-01533-9

Simulation of commercial fixed‑bed reactor for maleic


anhydride synthesis: application of different kinetic models
and industrial process data

Ivan Petric1 · Ervin Karić1

Received: 2 October 2018 / Accepted: 13 January 2019 / Published online: 23 January 2019
© Akadémiai Kiadó, Budapest, Hungary 2019

Abstract
In this study, we have developed a relatively simple mathematical model capable of
both simulating the synthesis of maleic anhydride from n-butane in industrial fixed-
bed reactor and determining the influences of inlet process parameters on reactor
performance. Ten kinetic models were used, and each of them was used in combina-
tion with a simplified reactor model. The validation of the developed mathemati-
cal model was performed using three process data sets with five process parameters
obtained from industrial fixed-bed reactor. The simulation results showed a good
agreement with the measured values for three kinetic models. The most influential
inlet process parameters are inlet flow rates of n-butane and oxygen. The maleic
anhydride yield is more sensitive to the changes of inlet process parameters than
n-butane conversion and maleic anhydride selectivity. By increasing the inlet
molar flow of n-butane for 20%, n-butane conversion and maleic anhydride yield
are decreased for 3.3% and 5.0%, while maleic anhydride selectivity is increased
for 6.7%. In order to increase the selectivity of maleic anhydride, it is necessary
to decrease the inlet temperature of reaction mixture, the inlet pressure of reaction
mixture as well as the inlet molar flow of oxygen, and to increase the inlet molar
flow of n-butane.

Keywords  Kinetic model · Reactor model · n-Butane · Maleic anhydride · Industrial


fixed-bed reactor · Simulation

Nomenclature
A Surface of heat exchange ­(m2)

Electronic supplementary material  The online version of this article (https​://doi.org/10.1007/s1114​


4-019-01533​-9) contains supplementary material, which is available to authorized users.

* Ivan Petric
ivan.petric@untz.ba
1
Department of Chemical Engineering, Faculty of Technology, University of Tuzla, 75000 Tuzla,
Bosnia and Herzegovina

13
Vol.:(0123456789)
1028 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

Ac Cross section of reactor tube ­(m2)


Ap External surface area of particle ­(m2)
a Area of heat exchange A per unit volume of reactor V ­(m−1)
CA Concentration of n-butane (kmol/m3)
CB Concentration of oxygen (kmol/m3)
CC Concentration of maleic anhydride (kmol/m3)
CD Concentration of carbon dioxide (kmol/m3)
CE Concentration of water (kmol/m3)
CF Concentration of carbon monoxide (kmol/m3)
CpA Specific heat capacity of n-butane (kJ/(kmol K))
CpB Specific heat capacity of oxygen (kJ/(kmol K))
CpC Specific heat capacity of maleic anhydride (kJ/(kmol K))
CpD Specific heat capacity of carbon dioxide (kJ/(kmol K))
CpE Specific heat capacity of water (kJ/(kmol K))
CpF Specific heat capacity of carbon monoxide (kJ/(kmol K))
Cpi Specific heat capacity of component i (kJ/(kmol K))
Ci Concentration of component i (kmol/m3)
CT0 Total inlet concentration of reaction mixture (kmol/m3)
Dp Effective diameter of particle in the bed (m)
du Inner diameter of reactor tube (mm)
Ea Activation energy (kJ/kmol)
FA Molar flow of n-butane (kmol/h)
FB Molar flow of oxygen (kmol/h)
FC Molar flow of maleic anhydride (kmol/h)
FD Molar flow of carbon dioxide (kmol/h)
FE Molar flow of water (kmol/h)
FF Molar flow of carbon monoxide (kmol/h)
Fi Molar flow of component i (kmol/h),
FT Total molar flow of reaction mixture (kmol/h)
FT0 Total molar flow of reaction mixture at reactor inlet (kmol/h)
FA0 Inlet molar flow of n-butane (kmol/h)
FB0 Inlet molar flow of oxygen (kmol/h)
G Superficial mass velocity (m/s)
k1 Rate constant in equations (12a) (mol L/(gcat h))
k2 Rate constant in equations (12b) (mol L/(gcat h))
k3 Rate constant in equations (12c) (mol L/(gcat h))
k1 Rate constant in equations (15a) (kmol/(kgca s Pa))
k2 Rate constant in equations (15b) (kmol/(kgcat s Pa))
k3 Rate constant in equations (15c) (kmol/(kgcat s Pa))
KB Adsorption equilibrium constant for n-butane (L/mol)
Kdiss Desorption constant ­(Pa−1)
k0 Pre-exponential factor (various units)
K1 Inhibition factor (–)
K2 Inhibition factor (–)
Ksorpt Sorption constant ­(Pa−1)
L Length of reactor tube (m)

13
Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054 1029

ṁ Mass flow of reaction mixture (kg/h)


Mi Molar mass of component i (kg/kmol)
Mmix Molar mass of mixture (kg/kmol)
pA Partial pressure of n-butane (atm)
pB Partial pressure of oxygen (atm)
pC Partial pressure of maleic anhydride (atm)
P Pressure of reaction mixture in reactor (Pa)
Pref Reference pressure (Pref = 101325 Pa)
P0 Inlet pressure of reaction mixture (Pa)
Pout Outlet pressure of reaction mixture (Pa)
R Universal gas constant (= 8.314 J/(mol K))
ri’ Reaction rate for component i (kmol/(kgcat s))
rji’ Rate of j reaction for i component (kmol/(kgcat s))
T Temperature of reaction mixture reactor (K)
Ta Ambient temperature (K)
Tref Reference temperature (Tref = 273.15 K)
Tout Outlet temperature of reaction mixture (K)
T0 Inlet temperature of reaction mixture (K)
U Overall heat transfer coefficient (kJ/(m2 h K))
V Volume of reactor ­(m3)
Vp Volume of particle ­(m3)
W Mass of catalyst (kg)
XA Conversion of n-butane (–)
yi Molar fraction of component i (–)
Ỹ C Overall yield of maleic anhydride (–)
S̃ C Overall selectivity of maleic anhydride (–)

Greek letters
α Pressure drop parameter ­(kg−1)
α1 Exponent in equation 16a (–)
α3 Exponent in equation 16c (–)
ΔHRj Heat of reaction j (kJ/kmol)
ΔHR1 Heat of the reaction 1 (kJ/kmol)
ΔHR2 Heat of the reaction 2 (kJ/kmol)
ΔHR3 Heat of the reaction 3 (kJ/kmol)
ε Porosity of catalyst (–)
μ Viscosity of gas mixture passing through the catalyst bed (kg/(m h))
υ Stoichiometric coefficient (–)
ρb Bulk density of catalyst bed (kg/m3)
ρc Density of catalyst particles (kg/m3)
ρ0 Density of reaction mixture at reactor inlet (kg/m3)
ϕA Outlet volume percentage of n-butane (V%)
ϕD Outlet volume percentage of carbon dioxide (V%)
ϕF Outlet volume percentage of carbon monoxide (V%)

13
1030 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

Introduction

Maleic anhydride is produced commercially by the selective catalytic oxidation


of n-butane over vanadium phosphorus oxide catalyst. This reaction is mainly
carried out in multi-tubular fixed-bed reactors filled with shaped catalyst parti-
cles. Such reactors can consist up to 30,000 tubes with diameter of 21–25  mm
and length of 3–6 m, Lohbeck et al. [1]. The main byproducts are CO and C ­ O 2,
as well as other byproducts like acrylic acid and acetic acid which are present
in minor amounts (Burnett et al. [2] Felthouse et al. [3]. Maleic anhydride is an
important intermediate in the industrial production of unsaturated resins, chemi-
cals for agriculture and food additives (Dente et al. [4]). The production capacity
of maleic anhydride will be increased world-wide by 30% in 2020 compared to
2015, Zhao [5]. Fixed-bed reactor is limited to operate with a low concentration
of n-butane (below 2.2%), while a fluidized-bed reactor allows to operate with a
higher concentration of n-butane, Cruz-López et  al. [6]. Maleic anhydride pro-
duction by selective oxidation of n-butane in a multi-tubular fixed-bed reactor
is constrained by the flammability limits. For modeling fixed-bed reactors, the
pseudo-homogeneous and heterogeneous models are the most popular ones due to
their accuracy and low computational cost, Jakobsen [7]. These models have been
applied to simulate n-butane oxidation in fixed-bed reactors (Ali and Humaizi
[8]; Brandstädter and Kraushaar-Czarnetzki [9]; Diedenhoven et al. [10]; Guettel
and Turek [11]; Sharma et  al. [12]) and membrane reactors (Alonso et  al. [13];
Marin et al. [14]). The phosphorus loss from a catalyst during long term opera-
tion has a strong effect on the temperature distribution in a reactor, which in turn
significantly affects the catalyst selectivity. Therefore, a reactor operation usually
requires the addition of ppm levels of an organic phosphorus compound together
with water to the reactor feed, Lesser et al. [15]. A packed bed catalytic reactor
is an assembly of usually uniformly sized catalytic particles, which are randomly
arranged and firmly held in position within a vessel or tube. The bulk fluid flows
through the voids of the bed. The reactants are transported first from the bulk of
the fluid to the catalyst surface, then through catalyst pores, where the reactants
adsorb on the surface of the pores and then undergo chemical transformation.
When the fluid containing the reactants flows through the packed bed a variety
of physical and chemical phenomena occur in the reactor. Due to the enormous
complexity of these phenomena an exact mathematical description of packed bed
reactors is virtually impossible, Iordanidis [16].
The developed mathematical model can be used to improve the understand-
ing of the process, to guide the choice of the optimal operation conditions, and
to suggest possible changes to the reactor design in order to improve the process
performance, Petric and Husanović [17]. The results of these studies can serve as
useful guidelines for improving the design and operation of the plants as well as
for improving the performance of the whole process without a need for expensive
tests on a plant.
Unfortunately, the validation of the developed models is rarely based on real
process data from industrial reactors. In most of the cases, researchers applied

13
Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054 1031

reactor and kinetic model to laboratory and pilot reactors instead of industrial
reactors. According to our knowledge based on an extensive review of the litera-
ture, this is the first study dealing with the comparison of different kinetic mod-
els for the synthesis of maleic anhydride from n-butane using the process data
from an industrial fixed-bed reactor. Therefore, the objectives of the present study
are the following: development of a simplified mathematical model for numeri-
cal simulation of partial oxidation of n-butane to maleic anhydride in industrial
fixed-bed reactor, validation of the model with measured parameters in industrial
fixed-bed reactor, as well as determining the influence of process parameters on
the reactor performance.

Mathematical model

Kinetic models

The mechanism of reaction set I has the following main reaction (Eq. 1) and side reac-
tions (Eqs. 2 and 3):
n − C4 H10 + 3.5O2 → C4 H2 O3 + 4H2 O (1)
n − C4 H10 + 6.5O2 → 4CO2 + 5H2 O (2)
C4 H2 O3 + 3O2 → 4CO2 + H2 O (3)
The mechanism of reaction set II has the following main reaction (Eq. 4) and side
reactions (Eqs. 5 and 6):
n − C4 H10 + 3.5O2 → C4 H2 O3 + 4H2 O (4)
n − C4 H10 + 4.5O2 → 4CO + 5H2 O (5)
n − C4 H10 + 6.5O2 → 4CO2 + 5H2 O (6)
The following nomenclature is used: A is n-C4H10, B is O ­ 2, C is C
­ 4H2O3, D is C
­ O2,
E is ­H2O, and F is CO.
The investigated kinetic models are: (7a–c), (8a–c) and (9a–c) (Alonso et al. [13]),
(10a–c) and (11a–c) (Buchanan and Sundaresan [18]), (12a–12c) (Centi et  al. [19]),
(13a–c) (Marin et al. [14]), (14a–c) (Lorences et al. [20]), (15a–15c) (Schneider et al.
[21]), and (16a–16c) (Sharma et al. [12]). The kinetic models are given by the follow-
ing equations:
[ Ea
] CA
r1� = k0 ⋅ e− R⋅T ⋅ [
(ja)
]
K1 ⋅CA K2 ⋅CC
1+ CB
+ CB

[ Ea
] CA
r2� = k0 ⋅ e− R⋅T ⋅ [
(jb)
]
K1 ⋅CA K2 ⋅CC
1+ CB
+ CB

13
1032 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

[ Ea
] CA
r3� = k0 ⋅ e− R⋅T ⋅ [
(jc)
]
K1 ⋅CA K2 ⋅CC
1+ CB
+ CB

Here j = 7, 8, 9, 10, 11, 13, 14.


k1 ⋅ KB ⋅ CA ⋅ CB𝛼
r1� = (12a)
1 + KB ⋅ CA

r2� = k2 ⋅ CB𝛽 (12b)

CB𝛾
( )
r3� = k3 ⋅ CC ⋅ (12c)
CA𝛿

( )1
K diss ⋅ p B
2

r1� = k1 ⋅ ( ) 1 ⋅ pA (15a)
1 + Kdiss ⋅ pB 2

Ksorpt ⋅ 10−5 ⋅ pB
r2� = k2 ⋅ ⋅ pA (15b)
1 + Ksorpt ⋅ 10−5 ⋅ pB

Ksorpt ⋅ 10−5 ⋅ pB
r3� = k3 ⋅ ⋅ pA (15c)
1 + Ksorpt ⋅ 10−5 ⋅ pB

{ Ea
}
𝛼
k0 ⋅ e− R⋅T ⋅ pA1
r1� = ( ) (16a)
1 + KII ⋅ pC

{ Ea
}
k0 ⋅ e− R⋅T ⋅ pC
r2� = ( )2 (16b)
1 + KII ⋅ pC

{ Ea
}
𝛼
r3� = k0 ⋅ e− R⋅T ⋅ pA3 (16c)

Here r1′ , r2′  , r3′ are reaction rates, R is universal gas constant (= 8.314 J/(mol K)), T
is the temperature of reaction mixture in a reactor, CA, CB, CC are concentrations
of n-butane, oxygen, and maleic anhydride and pA, pB, pC are partial pressures of
n-butane, oxygen, and maleic anhydride.

13
Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054 1033

The mechanism of reaction set I is used in the case of application of the kinet-
ics models (12a–12c) and (13a–c). The mechanism of reaction set II is used in the
case of application of the kinetics models (7a–c), (8a–c), (9a–c), (10a–c), (11a–c),
(14a–c), (15a–15c), and (16a–16c).
Kinetic parameters for kinetic models (7a–c), (8a–c), (9a–c) (Alonso et al. [13]),
(10a–c) and (11a–c) (Buchanan and Sundaresan [18]), (13a–c) (Marin et al. [14]),
(14a–c) (Lorences et  al. [20]), and (16a–16c) (Sharma et  al. [12]) are shown in
Table S1 (Supplementary material).
Kinetic parameters for kinetic model (12a–12c) (Centi et al. [19]) are shown in
Table S2 (Supplementary material).
Kinetic parameters for kinetic model (15a–15c) (Schneider et al. [21]) are shown
in Table S3 (Supplementary material).

Reactor model

The level of model detailedness is determined by the availability of accurate kinetic


expressions and model parameters. However, the applicability of the model can
also be limited by the difficulties associated with its mathematical handling. Equa-
tions encountered in packed bed reactor modeling range from algebraic equations
to multidimensional partial differential equations. Only very few and very simple
models permit exact analytical solutions. For the description of most chemical reac-
tors, we have to rely on simplified models capturing the most crucial and salient
features of the problem at hand. This also means that there is no universal model.
The best model is selected on the basis of the properties of the particular system
under consideration, the features of the system one is interested in, the availability
of the parameters included in the model and the prospects of successful numerical
treatment of the model equations, Iordanidis [16]. One-dimension and two-dimen-
sion 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, Quina et al. [22]. The
heterogeneous models are more difficult to apply because of the higher numeri-
cal requirements. Another disadvantage of the heterogeneous models is the higher
uncertainty in the heat and mass transfer properties and the heat conductivity of the
solids matrix, Brandstädter and Kraushaar-Czarnetzki [9].
The following assumptions and simplifications are used in development of reactor
model:

– there are no radial gradients of temperature and concentration in a reactor,


– reactor operates in a steady-state conditions,
– pseudo one-dimensional model is used,
– overall heat transfer coefficient has a constant value (taken from Diedenhoven
et al. [10]).

When there is not enough data available, once the model is proposed, the whole
effort has to be concentrated on a more precise estimation of the parameters in

13
1034 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

already known models, Martines et al. [23]. For the simulation of steady-state reac-
tor behavior at conditions not too close to runaway, a one-dimensional pseudo-
homogeneous plug-flow model generally is sufficient, whilst for a detailed design of
a tubular reactor two-dimensional heterogeneous reactor models with axial mass and
heat dispersion are usually applied, Koning [24]. Fixed-bed reactors for industrial
syntheses are generally operated in a stationary mode (i.e., under constant operat-
ing conditions) over prolonged production runs, and design therefore concentrates
on achieving an optimum stationary operation, Eigenberger [25]. The pseudo-homo-
geneous and heterogeneous models are the most popular ones for modeling fixed-
bed reactors due to their high accuracy and low computational cost, Jakobsen [7].
Sharma et al. [12] derived the overall heat transfer coefficient by asymptotic match-
ing of thermal fluxes between one-dimensional and two-dimensional models using
the effective radial thermal conductivity. Dixon [26] given the overall heat transfer
coefficient with relation which relates one- and two-dimensional fixed bed reactor
heat transfer models. For the purpose of this research, the value of overall heat trans-
fer coefficient has taken from literature, Diedenhoven et al. [10]. In the future study,
the calculation of overall heat transfer coefficient will be included in the model.
The molar balances of components are given by the following equation:
dFi
= ri� (17)
dW
Here i is component, Fi is molar flow of component i, ri′ is reaction rate for compo-
nent i and W is mass of catalyst.
The energy balance is given by the following equation:
U⋅a � � ∑3 � � �
⋅ T a − T + j=1
r ⋅ ΔH Rj
dT 𝜌b j
(18)
= ∑6
dW F ⋅C
i=1 i pi

Here i is number of component (i = 1,6), j is reaction (j = 1,3), U is overall heat trans-


fer coefficient, a is area of heat exchange A per unit volume of reactor V, ρb is bulk
density of catalyst bed, A is surface of heat exchange, V is volume of reactor, Cpi is
specific heat capacity of component i, Ta is ambient temperature, T is temperature of
reaction mixture in reactor, rj′ is rate of reaction j and ΔHRj is heat of reaction j.
The heat of reaction is given by the following equation:
ΔĤ R = a + b ⋅ T + c ⋅ T 2 + d ⋅ T 3 + e ⋅ T 4 (19)
Coefficients a, b, c, d and e for heat of reactions in Eq. (19) are shown in Table S4
(Supplementary material).
The specific heat capacity of component is given by the following equation:

Cp = a + b ⋅ T + c ⋅ 10−4 ⋅ T 2 + d ⋅ T 3 (20)
Coefficients a, b, c and d for specific heat capacities in the Eq. (19) are shown in
Table S5 (Supplementary material).
The concentrations of components are given by the following general equation:

13
Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054 1035

( ) ( ) ( )
Fi T0 P
Ci = CT0 ⋅ ⋅ ⋅ (21)
FT T P0
Here Ci is concentration of component i, CT0 is total inlet concentration of reaction
mixture, FT is total molar flow of reaction mixture, Fi is molar flow of component i,
T0 is inlet temperature of reaction mixture, T is temperature of reaction mixture in
reactor, P0 is inlet pressure of reaction mixture and P is pressure of reaction mixture
in reactor.
The total inlet concentration of reaction mixture CT0 is given by the following
equation:
P0
CT0 = (22)
R ⋅ T0
The total molar flow of reaction mixture FT is given by the following equation:
n

FT = Fi (23)
i=1

The relative rates of reaction in reaction j in compact notation are given by the
following equation:

rji� rjk�
= (24)
𝜐ji 𝜐jk

Here j is reaction, i and k are components, υ is stoichiometric coefficient and r


is reaction rate.
The pressure drop is given by the following equation:

dP 𝛼 T P F
=− ⋅ ⋅ ( 0) ⋅ T
dW 2 T0 P FT0 (25)
P0

Here α is pressure drop parameter. This parameter is given by the following


equation Fogler [27]:
[ ]
2 ⋅ G ⋅ (1 − 𝜀) 150 ⋅ (1 − 𝜀) ⋅ 𝜇
𝛼= ⋅ + 1.75 ⋅ G (26)
𝜌 0 ⋅ Dp ⋅ 𝜀 3 ⋅ 𝜌 b ⋅ A c ⋅ P 0 Dp

Here P0 is the pressure at reactor inlet, T0 is the inlet temperature of the reaction
mixture, T is the temperature of the reaction mixture in reactor, W is the mass of
catalyst, FT is the total molar flow of reaction mixture, FT0 is the total molar flow
of reaction mixture at reactor inlet, α is the pressure drop parameter, G is the super-
ficial mass velocity, εb is the porosity, Dp is the effective diameter of particle in the
bed, μ is the viscosity of gas mixture passing through the catalyst bed, ρ0 is the

13
1036 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

density of reaction mixture at reactor inlet, ρb is the density of catalyst bed, Ac is the
cross-sectional area.
Koning [24] calculated the pressure drop using the Brinkman or Ergun equation,
which were derived to describe the pressure drop in beds with a high aspect ratio, in
which wall channeling is not important.
The density of catalyst bed is given by the following equation:
W
𝜌b =
Ac ⋅ L (27)

Here L is the length of reactor tube, Ac is the cross section of reactor tube.
The density of catalyst particles is given by the following equation:
𝜌b
𝜌c = (28)
1−𝜀
Here ε is the porosity of catalyst (ε = 0.45).
The cross section of reactor tube is given by the following equation:
du2 ⋅ 𝜋
Ac = (29)
4
Here du is the inner diameter of reactor tube (du = 0.021 m).
The effective diameter of particle in the bed Dp is given by the following equation:

6 ⋅ Vp
Dp = (30)
Ap

Here Vp is the volume of particle, Ap is the external surface area of particle.


The superficial mass velocity G is given by the following equation:

G=
Ac (31)

Here ṁ is the mass flow of reaction mixture.


The viscosity of gas mixture passing through the catalyst bed μ, Šef and Olujić
[28] is given by the following equation:
∑n ∑n
𝜇 ⋅ yi ⋅ i=1 Mi
i=1 i
𝜇= ∑n ∑n (32)
y ⋅ i=1 Mi
i=1 i

Here μi is the viscosity of component i, yi is the molar fraction of component i, Mi is


the molar mass of component i.
The viscosity of each components in the gas mixture is calculated by the follow-
ing equation, based on data of pressure and temperature from the reference Green
and Perry [29]:
P 3600
( )
𝜇i = c1 + c2 ⋅ + c3 ⋅ (T − Tref ) ⋅ (33)
100 1 ⋅ 106

13
Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054 1037

The coefficients c1, c2 i c3 for different gas components can be found in Table  S6
(Supplementary material).
The density of reaction mixture at reactor inlet ρ0 is given by the following
equation:

Mmix Tref P0
𝜌o = ⋅ ⋅ (34)
22.4 T0 Pref

Here Mmix is the molar mass of mixture, Tref is the reference temperature
(Tref = 273.1 K), Pref is the reference pressure (Pref = 101325 Pa).
The molar mass of mixture is given by the following equation:
n

Mmix = yi ⋅ Mi (35)
i=1

The conversion of n-butane is given by the following equation:


FA0 − FA
XA = (36)
FA0
Here FA0 is inlet molar flow of n-butane and FA is molar flow of n-butane.
The overall yield of maleic anhydride is given by the following equation:
FC
Ỹ C = (37)
FA0
Here FC is molar flow of maleic anhydride.
The overall selectivity of maleic anhydride is given by the following equation:
FC
S̃ C = (38)
FA0 − FA
Fig. 1b shows the industrial reactor located in the Global Ispat Coke Industry
Lukavac. Maleic anhydride is produced with an annual output of 10,000 tons. The
synthesis is carried out in a fixed-bed reactor consisted of 11,600 tubes in a par-
allel arrangement with length (height) of 3.7 m, outside diameter of 25 mm and
wall thickness of 2 mm. The reactor tubes are filled up to height of 3.25 m with
total mass of 11.2 tons of catalyst POLYCAT MAC 4 ML (manufacturer Polynt,
Italy), based on vanadium-phosphorus oxide. The main characteristics of catalyst
particle are: shape = hollow cylinder, external diameter = 4.2  mm, inner diam-
eter = 1.2  mm, length = 4  mm, porosity of catalyst = 0.45 (Polynt, 2012). Other
design details of each tube can be found in Fig. 1a.
Input data for mathematical model are: T0 = 431.1 K, P0 = 134 kPa, FA0 = 20.98
kmol/h, FB0 = 262.1 kmol/h, FD0 = 0.4 kmol/h, FE0 = 6.72 kmol/h, Ta = 683.15 K,
W  = 0.966  kg, U  = 107  W/(m2 K) Diedenhoven et  al. [10], V  = 0.0013  m3, a 
=  0.26923 ­(m−1), R  = 8.314 J/(mol K).

13
1038 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

Tubesheet thickness 50 mm

Top spring 50 mm

Top
inert 125 mm

Total tube length 3700 mm Catalyst bed 3250 mm

Bottom
inert

Bottom spring 50 mm

Tubesheet thickness 50 mm

(a) (b)
Fig. 1  Industrial fixed-bed reactor: a schematic representation of one tube, b the outer surface of the
reactor

Numerical software package Polymath with Runge–Kutta–Fehlberg method was


used for a numerical solution of the set of differential equations.

Results and discussion

Tables 1 and 2 compare the simulation results with measured values of outlet pro-
cess parameters as well as percentage deviations of simulation results from meas-
ured values of outlet reactor process parameters. The measured outlet process
parameters are: temperature, pressure, volume percentages of n-butane, carbon diox-
ide and carbon monoxide. The best agreement of simulation results and measured
values was achieved with application of the kinetic models (12a–12c), (16a–16c)
and (15a–15c). The outlet temperatures of reaction mixture with application of the
kinetic models (12a–12c), (15a–15c), (16a–16c) were 677.8, 671.6, and 659.1  K,
while the outlet temperature of reaction mixture for industrial fixed-bed reactor in
Global Ispat Coke Industry Lukavac from December 2015, January 1016 and Febru-
ary 2016 were 682.7, 678.9, and 683.1 K.
The best agreement of simulation results and measured values for outlet tempera-
ture, outlet pressure, volume percentage of n-butane, outlet volume percentage of
carbon dioxide and outlet volume percentage of carbon monoxide, was achieved
with application of the kinetic model (12a–12c), the kinetic model (15a–15c), the
kinetic model (16a–16c), the kinetic models (16a–16c), the kinetic model (15a–15c)
and the kinetic model (16a–16c).

13
Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054 1039

Table 1  Comparisons of simulation results and measured values of outlet process parameters


Kinetic model Tout (K) Pout (bar) ϕA (V %) ϕD (V %) ϕF (V %)

A 682.7 0.664 0.29 1.05 1.03


B 678.9 0.672 0.28 1.08 1.06
C 683.1 0.662 0.30 1.30 1.06
(7a)–(7c) 640.6 0.650 0.62 3.65 0.53
(8a)–(8c) 639.9 0.662 0.59 8.60 9.24
(9a)–(9c) 633.6 0.728 1.95 7.02 0.21
(10a)–(10c) 651.0 1.114 1.40 2.27 0.71
(11a)–(11c) 630.6 0.683 0.63 1.53 0.10
(12a)–(12c) 677.8 0.672 0.43 0.96 –
(13a)–(13c) 655.6 0.760 1.21 3.64 –
(14a)–(14c) 631.3 1.116 0.64 4.33 2.18
(15a)–(15c) 671.6 0.663 0.31 1.26 1.40
(16a)–(16c) 659.1 0.654 0.30 1.19 1.40

A average measured values from December 2015, B average measured values from January 2016, C aver-
age measured values from February 2016

The largest deviations between model and industrial process data for the outlet
temperature of reaction mixture, the outlet pressure of reaction mixture, the outlet
volume percent of n-butane, the outlet volume percent of carbon dioxide and the
outlet percent of carbon monoxide, were observed with application of the kinetic
model (11a–c), the kinetic model (14a–c), the kinetic model (9a–c), the kinetic
model (8a–c) and the kinetic model (8a–c), respectively. Significant deviations
between model and industrial process data for outlet process reactor parameters were
also observed with application of kinetic models (7a–7c), (10a–10c) and (13a–13c).
Figs. S1–S10 (Supplementary material) compare the simulated and measured values
for temperatures of the reaction mixture along reactor length for different kinetic
models that were used in the simulation. There are several possible reasons for the
deviations between simulated and measured values for the above mentioned kinetic
models. Simplified reaction schemes may be one of the reasons because some side
reactions and formation of some by-products (such as acetic acid and acrylic acid)
have not been included in the studied mechanisms. Another possible reason for the
deviations may be relatively close values of activation energies and pre-exponen-
tial factors [for example, the kinetic model (8a–8c)], which leads to practically the
same temperature influence on the reaction rates for main reaction and side reactions
(Figs. S1–S5, S7–S8). The second part of these kinetic model equations contains the
concentrations of n-butane, oxygen and maleic anhydride. These concentrations are
interconnected so that any change in one concentration affects other concentrations.
Finally, the values of the kinetic parameters K1 and K2 are specific for the experi-
mental conditions in which the kinetic models are derived. Therefore, in order to
apply these models on industrial reactor, there is a need to optimize the values of the
kinetic parameters K1 and K2. In other words, the application of appropriate kinetic
data is crucial for reactor analysis and optimization.

13
1040 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

Table 2  Percentage deviations Kinetic model Tout (%) Pout (%) ϕA (%) ϕD (%) ϕF (%)
of simulation results from
measured values of outlet (7a)–(7c) 6.2 2.1 113.8 247.6 48.5
reactor process parameters
5.6 3.3 121.4 238.0 50.0
6.2 1.8 106.7 180.8 50.0
(8a)–(8c) 6.3 0.3 103.4 719.0 797.1
5.7 1.5 110.7 696.3 771.7
6.3 0.0 96.7 561.5 771.7
(9a)–(9c) 7.2 9.6 572.4 568.6 79.6
6.7 8.3 596.4 550.0 80.2
7.2 10.0 550.0 440.0 80.2
(10a)–(10c) 4.6 67.8 382.8 116.2 31.1
4.1 65.8 400.0 110.2 33.0
4.7 68.3 366.7 74.6 33.0
(11a)–(11c) 7.6 2.9 117.2 45.7 90.3
7.1 1.6 125.0 41.7 90.6
7.7 3.2 110.0 17.7 90.6
(12a)–(12c) 0.7 1.2 48.3 8.6 –
0.2 0.0 53.6 11.1 –
0.8 1.5 43.3 26.2 –
(13a)–(13c) 4.0 14.5 317.2 246.7 –
3.4 13.1 332.1 237.0 –
4.0 14.8 303.3 180.0 –
(14a)–(14c) 7.5 68.1 120.7 312.4 111.7
7.0 66.1 128.6 300.9 105.7
7.6 68.6 113.3 233.1 105.7
(15a)–(15c) 1.6 0.2 6.9 20.0 35.9
1.1 1.3 10.7 16.7 32.1
1.7 0.2 3.3 3.1 32.1
(16a)–(16c) 3.5 1.5 3.4 13.3 35.9
2.9 2.7 7.1 10.2 32.1
3.5 1.2 0.0 8.5 32.1

Table  S7 (Supplementary material) shows comparisons of the simulation


results of outlet process parameters for different kinetic models that were used
in the simulation. This mainly occurs since each kinetic model is based upon
a notably different range of experimental conditions, which points to the need
to use caution when using literature kinetics for purposes of industrial reactor
performance.
Alonso et  al. [13] studied kinetic models (7a–c), (8a–b) and (9a–c) and they
found that the possible reason for a poor agreement is a phenomenon of hot spots,
which is one of several factors constraining the operating range, pressure, tem-
perature, flow rate, and feed gas composition. Two other constraints include the
flammability limits and the maximum temperature.

13
Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054 1041

Buchanan and Sundaresan [18] studied the kinetic models (10a–c) and (11a–c)
for partial oxidation of n-butane over a vanadium phosphate catalyst. They also
measured and correlated the rates of steady-state n-butane oxidation over a range of
gas-phase compositions and temperatures, in order to assess the effects of phospho-
rus content on kinetics of n-butane and determine the effect of phosphorus on the
kinetics. The use of a vanadium phosphate catalyst with different ratios phospho-
rous/vanadium is also one of the reasons for a poor agreement of simulation results
and measured values (Fig. S4 and S5). On the other hand, the fixed-bed reactor in
Global Ispat Coke Industry Lukavac uses a vanadium-phosphorus oxide catalyst.
Buchanan and Sundaresan [18] used a reactor with internal tube diameter of 7 mm
and catalyst diameter of 4 mm. These values significantly differ from the values in
the fixed-bed reactor in Global Ispat Coke Industry Lukavac (internal diameter tube
of 21 mm, catalyst diameter of 2 mm) and this fact has a significant impact on agree-
ment of simulation results and measured values in the present study.
Centi et al. [19] investigated the kinetic model (12a–12c) and they used a fixed-
bed reactor based on a vanadium-phosphorus. Simulated temperatures of reaction
mixture along reactor length with the kinetic model (12a–12c) showed a good agree-
ment with measured values in the present study (Fig. S6), due to use of a similar
type of reactor and the same type of catalyst. A very good agreement was found
between calculated and measured temperature, as it has been shown in the study of
Romano et al. [30].
Marin et  al. [14] studied the kinetic model (13a–c) for partial oxidation of
n-butane to maleic anhydride in a membrane reactor with enhanced heat transfer
through the membrane walls. They also investigated the influences of reactor length,
flow rate of gas phase, inlet temperature of reaction mixture and inlet concentration
of n-butane on conversion of n-butane as well as selectivity of maleic anhydride.
They used fluidized-bed membrane reactor with the catalyst tubes (inner diameter of
34 mm, length of 0.5 m). In Global Ispat Coke Industry Lukavac, the reactor tubes
with inner diameter of 21 mm and length of 3.7 m are used. Therefore, the differ-
ences in the main dimensions of laboratory reactor versus industrial reactor are a
possible cause of a poor agreement of simulation results and measured values in the
present study (Fig. S7).
Lorences et al. [20] investigated the kinetic model (14a–c) with a wide range of
operating conditions in order to assess their impacts on environmental pollution,
selectivity of maleic anhydride, bio-based products, productivity and reaction rate.
The experiment was performed with a catalyst based on a vanadium phosphorus
oxide in a fluidized-bed reactor (inner diameter 0.04 m, height of 0.79 m). Volume
percentages of n-butane at reactor inlet were 2, 5 and 9%, Lorences et al. [20], while
in the industrial reactor of Global Ispat Coke Industry Lukavac, volume percentage
of n-butane at reactor inlet is 1.65%. Therefore, these differences in the inlet volume
percentages of n-butane could also be a reason for a poor agreement of simulation
and measured values in the present study (Fig. S8).
Schneider et  al. [21] investigated the kinetics (kinetic model (15a–15c)) of the
oxidation of n-butane over a catalyst based on a vanadium-phosphorus oxide using
the gradientless reactor with external recycle (reactor length of 6.5 m, internal diam-
eter of tube 1.15  cm). Simulation values of temperatures of the reaction mixture

13
1042 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

along reactor length for the kinetic model (Eqs. 15a–15c) show a good agreement
with the measured values in the present study (Fig. S9), due to the use of a similar
type of reactor and a similar type of catalyst.
Sharma et  al. [12] investigated the kinetic model (16a–16c) of selective oxida-
tion of n-butane to maleic anhydride. They used data from pilot fixed-bed reactor
(25 mm of outer diameter, 5 m long) with a catalyst based on a vanadium-phospho-
rus oxide. Their temperature profile along the reactor length was well predicted. The
application of the kinetic model (16a–16c) on industrial process data in the present
study also resulted in a good prediction of the temperature profile along the reactor
length (Fig. S10). This is probably due to the application of the same type of reactor
and the similar type of catalyst. Table 3 compares operating and experimental/pre-
dicted conditions from the study of Sharma et al. [12] and the present study.
Diedenhoven et al. [10] developed a model for the phosphorus dynamics in vana-
dium phosphorus oxide (VPO) catalysts for the oxidation of n-butane to maleic
anhydride. If no phosphorus is added to the reactant feed, the catalytic activity
increases until runaway occurs. With addition of a proper amount of phosphorus, the
loss can be compensated while excessive phosphorus addition results in complete
catalyst deactivation. Changing operation conditions with respect to reactant con-
centrations and temperature can damage the V-P-0 catalyst irreversibly, Schneider
et al. [21].
Figs. 2, 3 and 4 show the conversion of n-butane, the yield of maleic anhydride,
the selectivity of maleic anhydride, the molar flow rate of n-butane, the molar flow
rate of maleic anhydride and the molar flow rate of oxygen, all along reactor length
(Eqs. 16a–16c).
Centi et al. [19] investigated the dependence of the conversion of n-butane as a
function of space time. For higher conversions, the experimental yield of maleic
anhydride is lower than the calculated value. Therefore, a successive reaction of
maleic anhydride combustion must occur. A strong increase in the yields of both
carbon monoxide and carbon dioxide was found, occurring simultaneously with a
decrease of the yield of maleic anhydride.
Romano et  al. [30] showed that the conversion of n-butane is increased along
the reactor length for a fixed-bed reactor. Fluid-dynamic and chemical models of
an industrial-scale turbulent fluidized-bed reactor for partial oxidation of n-butane
to maleic anhydride catalyzed by (VO)2P2O7 solid particles were proposed. The
normalized conversion of n-butane as a function of a dimensionless reactor height
was investigated. It was shown that the normalized conversion of n-butane increased
along the reactor height (length), which the present study has also confirmed
(Fig.  2). Due to the rising catalytic activity, the conversion of n-butane increases
as a result of the increase in space time. Varma and Saraf [31] showed that the con-
version calculated with pseudo-homogeneous model has excellent agreement with
the experimental data. Similar conclusions may be drawn for the yield of maleic
anhydride.
Alonso et al. [13] investigated the yield of maleic anhydride along the length
of reactor tube. They found that the yield of maleic anhydride is increased along
the reactor length, which has been shown in the present study (Fig. 3). Tempera-
ture rises gradually and approaches to the maximum value and then it remains

13
Table 3  Comparison of operating and experimental/predicted conditions from the study of Sharma et al. (1991) and the present study
Operating conditions Experimental/predicted conditions Reference
3
Feed rate ­(m /h) Inlet butane Salt tempera- Inlet pres- Hot spot (°C) Conversion (%) Selectivity (%) Pressure drop (atm)
conc. (%) ture (°C) sure (atm)

1.68 1.86 390 1.65 419/424 84/77.2 68/63.4 065/0.61 Sharma et al. [12]
Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

1.34 1.73 408 1.34 430/415 /90 /89 0.69/0.68 Present study
1043

13
1044 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

Fig. 2  Conversion of n-butane along reactor length

Fig. 3  Yield of maleic anhydride along reactor length

constant, which is a quite different from the classical fixed-bed hot spot pattern.
This may imply that longer tubes could improve maleic anhydride production
because the concentration of maleic anhydride is increased almost linearly along
the tube length. The oxygen concentration is also increased along the tube length
and its exit concentration could be a limiting factor.
Maria and Dan [32] showed that both the conversion of n-butane and the yield
of maleic anhydride are increased along the reactor length for a fixed-bed reactor,
which has been shown in the present study as well (Figs. 2 and 3). The concentra-
tions of maleic anhydride, carbon monoxide, and carbon dioxide are increased
monotonically along the reactor length, Burnett et  al. [2], which corresponds to
the findings of the present study.

13
Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054 1045

Fig. 4  Selectivity of maleic anhydride along reactor length

Moser and Schrader [33] studies the selective oxidation of n-butane to maleic
anhydride using two vanadium-phosphorus-oxygen catalysts, β-VOP04 and
(VO)2P2O7. The selectivity of maleic anhydride as a function of space time was also
investigated. For β-VOP04 catalyst the selectivity of maleic anhydride increased
with the increase of space time (which is directly proportional to volume and length
of a reactor), which has been shown in the present study (Fig.  4). In the study of
Moser and Schrader [33], for (VO)2P2O7 catalyst the selectivity of maleic anhydride
first increased and then, after certain time, decreased. On the other hand, the pre-
sent study has demonstrated the continuous increase of the selectivity. Therefore, the
choice of a catalyst is very important for the selectivity of maleic anhydride.
Operation at the maximum yield is, in general, preferred. Operation at conversion
below the optimum results in lowered yield because too much unconverted n-butane
passes through the reactor. Operation at conversions higher than the optimum
results in lowered yields because increasingly higher levels of maleic anhydride are
degraded to CO and ­CO2, Burnett et al. [2].
Tables 4, 5, 6 and 7 show the influences of inlet process parameters on the per-
formance of synthesis of maleic anhydride from n-butane in industrial fixed-bed
reactor.
By increasing the inlet temperature of reaction mixture, the conversion of
n-butane is increased while both the selectivity of maleic anhydride and the yield
of maleic anhydride are decreased. By increasing the inlet temperature of reac-
tion mixture for 6  °C, the conversion of n-butane is increased for 4.4%, the yield
of maleic anhydride is decreased for 3.8%, while the selectivity of maleic anhy-
dride is decreased for 4.5%. Decreasing the inlet temperature of reaction mixture
has the opposite effect on the simulated results. By decreasing the inlet temperature
of reaction mixture for 6 °C, the conversion rate is decreased for 7.8%, the yield of
maleic anhydride is increased for 8.8%, while the selectivity of maleic anhydride is
increased for 5.6%. The selectivity of maleic anhydride exhibits the typical opposite
tendency compared to the conversion of n-butane.

13
1046 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

Table 4  Influence of inlet T0 (K) XA (–) Ỹ C (–) S̃ C (-–)


temperature of reaction mixture
(T0) on conversion of n-butane
425.1 0.83 0.87 0.94
(XA), yield ( Ỹ C ), and selectivity
of maleic anhydride ( S̃ C) 427.1 0.86 0.84 0.93
429.1 0.86 0.84 0.92
431.1 0.90 0.80 0.89
433.1 0.91 0.78 0.87
435.1 0.93 0.77 0.86
437.1 0.94 0.77 0.85

Table 5  The influence of inlet P0 (bar) XA (–) Ỹ C (–) S̃ C (–)


pressure of reaction mixture (P0)
on conversion of n-butane (XA),
1.28 0.94 0.73 0.93
yield ( Ỹ C ), and selectivity of
maleic anhydride ( S̃ C) 1.30 0.92 0.77 0.93
1.32 0.91 0.79 0.91
1.34 0.90 0.80 0.89
1.36 0.89 0.80 0.88
1.38 0.87 0.81 0.87
1.40 0.87 0.83 0.86

Table 6  Influence of inlet FA0 (kmol/h) XA (–) Ỹ C (–) S̃ C (–)


molar flow of n-butane (FA0)
on conversion of n-butane (XA),
− 20 0.97 0.89 0.85
yield ( Ỹ C ), and selectivity of
maleic anhydride ( S̃ C) − 15 0.94 0.87 0.87
− 10 0.93 0.84 0.87
− 5 0.92 0.83 0.89
0 0.90 0.80 0.89
+ 5 0.89 0.78 0.92
+ 10 0.89 0.78 0.92
+ 15 0.88 0.77 0.94
+ 20 0.87 0.76 0.95

Higher temperatures lead to faster combustion reactions, which leads to the


formation of temperature gradients that affect the reduction of selectivity. Parto-
pour and Dixon [34] observed loss of selectivity and yield in the center of the bed
at increased bed depths, because of the temperature increase and decrease in gas
phase oxygen there.
Ballarini et  al. [35] observed a lower maleic anhydride selectivity at higher
temperatures which was partly due to the formation of higher molecular weight
compounds at total oxygen conversions.

13
Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054 1047

Table 7  Influence of inlet FB0 (kmol/h) XA (–) Ỹ C (–) S̃ C (–)


molar flow of oxygen (FB0) on
conversion of n-butane (XA),
− 20 0.95 0.86 0.92
yield ( Ỹ C ), and selectivity of
maleic anhydride ( S̃ C) − 15 0.93 0.84 0.90
− 10 0.91 0.81 0.90
− 5 0.91 0.81 0.89
0 0.90 0.80 0.89
+5 0.88 0.78 0.85
+ 10 0.88 0.76 0.85
+ 15 0.87 0.74 0.84
+ 20 0.87 0.73 0.83

Since the activation energy of the first reaction is lower than the activation energy
of the second reaction, reactions should be carried out at low temperatures to maxi-
mize the selectivity of maleic anhydride. In other words, the lower temperature is
more favorable for the reaction with lower activation energy.
The inlet gas temperature has a strong influence on behavior of the reactor, due to
the exponential influence on reaction kinetics. As expected, by increasing the inlet
temperature, butane conversion increases due to the increasing reaction rates, while
selectivity decreases.
Centi et al. [19] showed that by increasing the inlet temperature of reaction mix-
ture, the conversion of n-butane is increased, but the selectivity of maleic anhydride
is decreased. Similar results are obtained in the present study.
Lorences et al. [20] showed that by increasing the inlet temperature of reaction
mixture from 623 to 653 K for different inlet volume percent of n-butane, the con-
version of n-butane is increased from 8.04 to 43.08%, while the selectivity of maleic
anhydride is decreased from 64.61 to 56.65%. Similar trend is observed in the pre-
sent study.
Varma and Saraf [31] showed that by increasing the inlet temperature of reaction
mixture the yield of maleic anhydride is increased, which is in opposition with the
results in the present study.
Ghaznavi et al. [36] showed that by increasing the inlet temperature of reaction
mixture the selectivity of maleic anhydride is decreased, which has been shown in
the present study.
Patience and Bockrath [37] showed that selectivity increased with decreas-
ing temperature and approached 80% at 360 °C. Therefore, high reaction tempera-
ture has negative effect on the selectivity of maleic anhydride, which leads to its
decreasing.
When the temperature was raised within the 400–435  °C interval, the n-butane
conversion increased, while the selectivity of maleic anhydride decreased, Dente
et al. [4]. This is consistent with the well-known detrimental effect of high reaction
temperatures on the selectivity of maleic anhydride. Consequently, the selectivity of
maleic anhydride is more favorable for low temperatures, while ­CO2 selectivity will
be higher at high temperatures.

13
1048 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

Parallel reactions of n-butane combustion and oxidative degradation to ace-


tic and acrylic acids impair useful yields, Trifirò and Grasselli [38]. In some
future work, these parallel reactions of n-butane combustion should be taken into
account. Higher inlet temperatures generate higher hot spots, which affect reactor
performance. The strong selectivity decrease at low conversions can be assigned
to the high hot-spot temperatures near the inlet of the reactor, where the conver-
sion is still low, Guettel and Turek [11]. Development of hot spots has a negative
effect on the activity of catalyst, which is one of the reasons for reduced yield of
maleic anhydride. It is very difficult to control the temperature in industrial fixed
bed reactor because of highly exothermic reactions and a large number of tubes
(11,600 tubes in this case).
Due to the rising catalytic activity, the conversion of n-butane increases as a
function of time while the maleic anhydride selectivity decreases, Diedenhoven
et al. [10]. These findings are in agreement with results of the present study.
Catalyst deactivation was not taken into account in this study. It should be
noted that process data from industrial reactor were obtained with using a fresh
catalyst.
By increasing the inlet pressure of reaction mixture, the conversion of n-butane,
and selectivity of maleic anhydride are decreased, while the yield of maleic anhy-
dride is increased. These findings are in agreement with the results of the Dente
et al. [4] as they showed that by increasing the pressure of reaction mixture at tem-
perature of 410 °C and n-butane concentration of 1.6 V %, yields of maleic anhy-
dride, carbon monoxide, carbon dioxide, acetate and acrylic acid are increased.
By increasing the inlet pressure of the reaction mixture for 0.06 bar, conversion of
n-butane is decreased for 3.3%, yield of maleic anhydride is increased for 3.8%,
while selectivity of maleic anhydride is decreased for 3.4%. Decreasing the inlet
pressure of reaction mixture has the opposite effect on the simulated results. By
decreasing the inlet pressure of reaction mixture for 0.06 bar, conversion of n-butane
increased for 4.4%, yield of maleic anhydride is decreased for 8.8%, while the selec-
tivity of maleic anhydride is increased for 4.5%.
Wellauer et al. [39] showed that by increasing the inlet pressure of reaction mix-
ture, the selectivity of maleic anhydride is increased, which is in opposition with
findings in the present study. The development of temperature hot-spots has a detri-
mental effect on reactor yield and catalyst lifetime. If these hot-spots were avoided
and if the reaction was carried out in an isothermal reactor, yield and catalyst life-
time would be improved.
Cruz-López et al. [6] found that the selectivity decreases when changing butane
partial pressure from 12 to 220 hPa. This is confirmed by the results of the present
study.
Although the butane conversion decreased with increasing butane concentration,
the overall consumption rate of n-butane was enhanced by the higher butane concen-
tration since more oxygen reacted at the higher butane concentration. With increas-
ing butane concentration, the selectivity of maleic anhydride slightly declined but
the selectivity to ­COx gradually increased.
By increasing the butane concentration, the maleic anhydride yield could be
increased substantially, Hutchings [40].

13
Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054 1049

Under high n-butane inlet feed concentrations, maleic anhydride selectivity is


lower when the oxygen conversion is near complete due to higher carbon oxides and
tetrahydrophthalic and phthalic anhydrides.
Maleic anhydride production rates increase even with a feed composition con-
taining 1.4 V% n-butane in air. A near equimolar feed of 6% n-butane and oxygen
resulted in the highest maleic anhydride production rate, Shekari and Patience [41].
The overall conversion of n-butane decreases with an increase in the butane feed
concentration, since the loss of n-butane due to undesired side reactions is also
reduced.
For an increase in n-butane feed concentration, the reaction rate does not increase
proportionally, and thus conversion is decreased, Pugsley et al. [42].
At low reactor pressures, the pressure drop over the reactor was significant (up to
18% at 2 bar), Koning [24]. At a constant mass flow of the feed, the pressure drop
over the reactor increased when lowering the reactor pressure. At the highest flow
rate, the measured pressure drop was less than 0.25 bar for experiments at pressures
of 4 bar and higher and was maximum 0.4 bar at the lowest reactor pressure of 2 bar,
Koning [24]. In the present study, the calculated pressure drop is 0.7 bar.
With increasing macro-pore porosity, molecules can diffuse faster in and out of
the catalyst pellet and hence the conversion will increase at first. At some value of
macro-pore porosity (between 0.2 and 0.25) the loss of active surface becomes too
high and conversion decreases. The decrease of the selectivity is probably because
the parallel and consecutive reactions are both enhanced with higher surface area
and slower diffusion, Dong et al. [43].
Shekari and Patience [44] showed that higher pressures improved the yield of
maleic anhydride, by maintaining the catalyst in a higher oxidation state and thus
increasing catalytic activity, which more than compensated for the decrease in
maleic anhydride selectivity. This is confirmed by the results of the present study.
High pressure also affects the redox reaction rates and activation energies. With
increasing the pressure, n-butane conversion increased by about 40–60%, Shekari
and Patience [44]. High pressure maintains the catalyst in a higher oxidation state
(as long as there is sufficient oxygen in the gas phase) and as a consequence, the
catalytic activity is improved together with maleic anhydride yield, Shekari and
Patience [44]. This may be one of the reasons for the increase of maleic anhydride
yield, as shown in this study.
By increasing the inlet molar flow of n-butane, the conversion of n-butane
is decreased, the yield of maleic anhydride is decreased, while the selectiv-
ity of maleic anhydride is increased. Because the partial oxidation of n-butane
to maleic anhydride is an exothermic reaction, the generated heat can be trans-
ferred faster through a reactor at higher molar flow of n-butane. The increase of
n-butane molar flow could also make the reaction products leave the catalyst bed
in time to avoid deep oxidation. Therefore, the increase of molar flow of n-butane
is advantageous to the maleic anhydride selectivity but it is disadvantageous
to the n-butane conversion. By increasing the reactor length, butane conver-
sion increases due to the increase in the residence time. By increasing the inlet
molar flow of n-butane for 20%, conversion of n-butane is decreased for 3.3%, the
yield of maleic anhydride is decreased for 5.0%, while the selectivity of maleic

13
1050 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

anhydride is increased for 6.7%. Decreasing the inlet molar flow of n-butane has
the opposite effect on the simulated results. By decreasing the inlet molar flow of
n-butane for 20% the conversion rate of n-butane is increased for 7.8%, the yield
of maleic anhydride is increased for 11.3%, while the selectivity of maleic anhy-
dride is decreased for 4.5%. The increase of the inlet molar flow rate of n-butane
leads to the increase of the selectivity of maleic anhydride, Fernández et al. [45],
which has been shown in the present study.
Hongbing and Lefu [46] showed that by increasing the inlet molar flow of
n-butane, the conversion of n-butane is decreased, the yield of maleic anhydride is
decreased, while the selectivity of maleic anhydride is increased.
Although the yield of maleic anhydride decreases with increasing molar flow rate
of n-butane, the number of moles of maleic anhydride formed, and hence the prod-
uct concentration at the reactor exit increases, resulting in higher space–time yields
and a more profitable process, Emig et al. [47].
Increasing the inlet molar flow rate of n-butane results in decreasing the conver-
sion of n-butane and increasing the selectivity of maleic anhydride. As a result, both
the yield and production of maleic anhydride also decrease, Fernández et al. [45].
Maintaining sufficient oxygen together with n-butane is critical to maintaining
high maleic anhydride productivity. This observation could be attributed to the
opposing positive and negative effects of elevated n-butane concentration in the feed
to the reactor. The positive effect is related to the increased reaction rate with high
n-butane concentrations. However, high n-butane concentrations negatively affect
the catalyst performance by decreasing the catalyst oxidation state and by accelerat-
ing catalyst deactivation (loss of surface oxygen) or probably due to higher rate of
surface carbon formation.
A decrease in maleic anhydride selectivity is accompanied by an increase of the
selectivities to CO and ­CO2, Hofmann et al. [48].
By increasing the inlet molar flow of oxygen the conversion of n-butane is
decreased, the yield of maleic anhydride is decreased, and the selectivity of maleic
anhydride is also decreased. By increasing the inlet molar flow of oxygen for 20%,
conversion of n-butane is decreased for 3.3%, the yield of maleic anhydride is
decreased for 8.7%, while the selectivity of maleic anhydride is decreased for 6.7%.
Decreasing the inlet molar flow of oxygen has the opposite effect on the simulated
results. By decreasing the inlet molar flow of oxygen for 20%, the conversion of
n-butane is increased for 5.5%, the yield of maleic anhydride is increased for 7.5%,
while the selectivity of maleic anhydride is increased for 3.3%.
Fernández et  al. [45] studied the effect of different oxygen feeding scenarios
in a fixed-bed reactor for the production of maleic anhydride. The reactor perfor-
mance in terms of the produced maleic anhydride mole fraction can be improved by
decreasing the oxygen ratio in the reactor single feed by 70%. Variations of oxygen
to butane ratios along the reactor could affect the overall performance of the reactor.
The kinetics of the desired and undesired reactions will result in either higher or
lower selectivity or yield of the desired product due to the variable oxygen concen-
tration, Ali et al. [49].
Contractor et al. [50] showed that increased conversion of n-butane and decreased
selectivity of maleic anhydride are obtained with increasing the inlet molar flow of

13
Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054 1051

oxygen. However, adding more oxygen in the reactor feed decreases the selectivity
of maleic anhydride, which has been shown in the present study.
By increasing the inlet molar flow of oxygen, the conversion of n-butane is
decreased, the yield of maleic anhydride is decreased, Hongbing and Lefu [46].
These findings are in accordance with results of the present study.
At high oxygen/n-butane ratios (more than the optimum value), the catalyst is too
oxidized and therefore less selective, Gascón et al. [51].
The maximum yield of maleic anhydride is limited by the following factors, Bal-
larini et al. [35]: (a) the presence of parallel reactions of n-butane combustion and of
oxidative degradation to acetic and acrylic acids (these are characterized by higher
activation energies with respect to the main reaction); (b) the presence of consecu-
tive reactions of combustion, which lower the selectivity to maleic anhydride when
the alkane conversion is increased. If no phosphorus is added to the reactant feed,
the catalyst loses phosphorus with time on stream accompanied by an increase in
activity and a decrease in maleic anhydride selectivity, Diedenhoven et  al. [10].
Increasing the inlet molar flows of oxygen without adding phosphorus may be one
of the reasons for reduced selectivity.
The results of the present study have demonstrated the need for the improvement
of the existing kinetic models for partial oxidation of n-butane to maleic anhydride
by determination of the optimized kinetic parameters. In order to accomplish this
task, additional data sets of process variables should be provided. In this way, better
agreement between simulated and measured values of industrial fixed-bed reactor
can be expected in some future studies. Another possible research direction would
be modification of a simplified reactor model used in the present study by introduc-
ing mass and heat transfer in a two-dimensional reactor model. Finally, the most
important task is to optimize the process performance in industrial fixed-bed reactor.
The results of the present study indicate the key inlet process parameters (tempera-
ture of reaction mixture, pressure of reaction mixture, molar flow of n-butane, molar
flow of oxygen), but a real challenge would be a simultaneous process optimization
(i.e. determination of the optimal values of these parameters).

Conclusions

This study showed that the existing ten kinetic models for partial oxidation of
n-butane to maleic anhydride have certain limitations concerning the process behav-
ior predictions in a commercial fixed-bed reactor. Therefore, these kinetic models
need to be modified in order to achieve better agreement between simulation results
of the main process parameters and industrial process data. This modification
implies the evaluation of the optimal values of kinetic parameters on the basis of
process data from an industrial reactor. In fact, the application of three kinetic mod-
els showed a good agreement with the main process parameters but none of these
models accurately predicted all of the main process parameters. The most influential
inlet process parameters are flow rate of n-butane and flow rate of oxygen. The yield
of maleic anhydride is more sensitive to the changes of inlet process parameters than
conversion of n-butane and selectivity of maleic anhydride. In order to increase the

13
1052 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

selectivity of maleic anhydride, it is necessary to decrease the inlet temperature of


reaction mixture, the inlet pressure of reaction mixture as well as the inlet molar
flow of oxygen, and to increase the inlet molar flow of n-butane. The future work
is directed to improvement of the mathematical model (both reactor and kinetic
model), validation of the improved mathematical model and a rigorous optimization
of the process performance.

Acknowledgements  The research conducted and presented within the study was a part of research pro-
ject, financially supported by the Federal Ministry of Education and Science of Bosnia and Herzegovina
(05-39-2482-1/17). The authors would like to thank Technical Manager of Maleic Anhydride Plant Mr.
Ermin Mujkić, for providing process data.

References
1. Lohbeck K, Haferkorn H, Fuhrmann W, Fedtke N (2000) Maleic and fumaric acids. Ullmann’s
encyclopedia of industrial chemistry. Wiley, New York
2. Burnett JC, Keppel RA, Robinson WD (1987) Commercial production of maleic anhydride by cata-
lytic processes using fixed bed reactors. Catal Today 1:537–586
3. Felthouse TR, Burnett JC, Horrel B, Mummey MJ, Kuo Y (2001) (200) Maleic anhydride, maleic
acid, and fumaric acid. Kirk–Othmer encyclopedia of chemical technology. Wiley, New York
4. Dente M, Pierucci S, Tronconi E, Cecchini M, Ghelfi F (2003) Selective oxidation of n-butane to
maleic anhydride in fluid bed reactors: detailed kinetic investigation and reactor modelling. Chem
Eng Sci 58:643–648
5. Zhao A (2001) Maleic anhydride chain—world market overview. In: Asian Petrochemical Industry
Conference, Korea Petrochemical Industry Association. Seoul, Republic of Korea
6. Cruz-López A, Guilhaume N, Miachon S, Dalmon JA (2005) Selective oxidation of butane to maleic
anhydride in a catalytic membrane reactor adapted to rich butane feed. Catal Today 107:949–956
7. Jakobsen HA (2014) Chemical reactor modeling: multiphase reactive flows. Springer Science &
Business Media, Berlin
8. Ali MAH, Al-Humaizi K (2014) Maleic anhydride production in a cross-flow reactor: a comparative
study. Can J Chem Eng 92:876–883
9. Brandstädter WM, Kraushaar-Czarnetzki B (2007) Maleic anhydride from mixtures of n-butenes
and n-butane: simulation of a production-scale nonisothermal fixed-bed reactor. Ind Eng Chem Res
46:1475–1484
10. Diedenhoven J, Reitzmann A, Mestl G, Turek T (2012) A model for the phosphorus dynamics of
VPO catalysts during the selective oxidation of n-butane to maleic anhydride in a tubular reactor.
Chem Ing Tech 84:517–523
11. Guettel R, Turek T (2010) Assessment of micro-structured fixed-bed reactors for highly exothermic
gas-phase reactions. Chem Eng Sci 65:1644–1654
12. Sharma RK, Cresswell DL, Newson EJ (1991) Kinetics and fixed-bed reactor modeling of butane
oxidation to maleic anhydride. AIChE 37:39–47
13. Alonso M, Lorences MJ, Pina MP, Patience GS (2001) Butane partial oxidation in an externally
fluidized bed-membrane reactor. Catal Today 67:151–157
14. Marin P, Hamel C, Ordonez S, Diez FV, Tsotsas E, Seidel-Morgenstern A (2010) Analysis of a flu-
idized bed membrane reactor for butane partial oxidation to maleic anhydride: 2D modelling. Chem
Eng Sci 65:3538–3548
15. Lesser D, Mestl G, Turek T (2016) Transient behavior of vanadyl pyrophosphate catalysts during
the partial oxidation of n-butane in industrial-sized, fixed bed reactors. Appl Catal A 510:1–10
16. Iordanidis AA (2002) Mathematical modeling of catalytic fixed bed reactors. Twente University
Press, Enschede
17. Petric I, Husanović A (2015) Comparison of different kinetic models for the chlorohydrin process
using real data from an industrial tubular reactor. Can J Chem Eng 93:78–87
18. Buchanan JS, Sundaresan S (1986) Kinetics and redox properties of vanadium phosphate catalysts
for butane oxidation. Appl Catal 26:211–226

13
Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054 1053

19. Centi G, Fornasari G, Trifiro F (1985) n-Butane oxidation to maleic anhydride on vanadium-
phosphorus oxides: kinetic analysis with a tubular flow stacked-pellet reactor. Ind Eng Chem Res
24:32–37
20. Lorences MJ, Patience GS, Díez FV, Coca J (2003) Butane oxidation to maleic anhydride: kinetic
modeling and byproducts. Ind Eng Chem Res 42:6730–6742
21. Schneider P, Emig G, Hofmann H (1987) Kinetic investigation and reactor simulation for the cata-
lytic gas-phase oxidation of n-butane to maleic anhydride. Ind Eng Chem Res 26:2236–2241
22. Quina MJ, Almeida-Costa AC, Quinta-Ferreira RM (2007) Fixed-bed reactor modeling and sim-
ulation with e-learning tools. In: International Conference on Engineering Education, Coimbra,
Portugal
23. Martinez OM, Duarte SP, Lemcoff NO (1985) Modeling of fixed bed catalytic reactors. Comp
Chem Eng 9:535–545
24. Koning GW (2002) Heat and mass transport in tubular packed bed reactors at reacting and non-
reacting conditions: experiments and models. Twente University Press, Enschede
25. Eigenberger G (1992) Fixed-Bed Reactors. Vol. B4 of Ullmann’s Encyclopedia of Industrial Chem-
istry. VCH VerJagsgesellschaft. Weinheim, Germany
26. Dixon AG (1985) Thermal resistance models of packed-bed effective heat transfer parameters.
AIChE 31:826–834
27. Fogler HS (2005) Elements of chemical reaction engineering, 4th edn. Prentice Hall, New Jersey
28. Šef F, Olujić Ž (1988) Projektiranje procesnih postrojenja. Savez kemičara i tehnologa Hrvatske.
Kemija u industriji, Zagreb, Hrvatska
29. Green DW, Perry RH (2007) Perry’s chemical engineers’ handbook, 8th edn. McGraw-Hill Educa-
tion, New York
30. Romano A, Di Giuliano A, Gallucci K, Foscolo PU, Cortelli C, Gori S, Novelli M (2016) Simula-
tion of an industrial turbulent fluidized bed reactor for n-butane partial oxidation to maleic anhy-
dride. Chem Eng Res Des 114:79–88
31. Varma RL, Saraf DN (1978) Oxidation of butene to maleic anhydride: II. Effect of physical trans-
port on reactor performance. J Catal 55:373–383
32. Maria G, Dan A (2012) Setting optimal operating conditions for a catalytic reactor for butane oxida-
tion using parametric sensitivity analysis and failure probability indices. J Loss Prev Process Ind
25:1033–1043
33. Moser TP, Schrader GL (1985) Selective oxidation of n-butane to maleic anhydride by model VPO
catalysts. J Catal 92:216–231
34. Partopour B, Dixon AG (2018) N-butane partial oxidation in a fixed bed: a resolved particle compu-
tational fluid dynamics simulation. Can J Chem Eng 96:1946–1956
35. Ballarini N, Cavani F, Cortelli C, Gasparini F, Mignani A, Pierelli F, Mazzoni G (2005) The con-
tribution of homogeneous and non-oxidative side reactions in the performance of vanadyl pyroph-
osphate, catalyst for the oxidation of n-butane to maleic anhydride, under hydrocarbon-rich condi-
tions. Cataly Today 99(1–2):115–122
36. Ghaznavi T, Neagoe C, Patience GS (2014) Partial oxidation of D-xylose to maleic anhydride and
acrylic acid over vanadyl pyrophosphate. Biomass Bioenerg 71:285–293
37. Patience GS, Bockrath RE (2010) Butane oxidation process development in a circulating fluidized
bed. Appl Catal A 376(1–2):4–12
38. Triffiro F, Grasselli RK (2014) How the yield of maleic anhydride in n-butane oxidation, using VPO
catalysts, was improved over the years. Top Catal 57(14–16):1188–1195
39. Wellauer TP, Cresswell DL, Newson EJ (1986) Optimal policies in maleic anhydride production
through detailed reactor modelling. Chem Eng Sci 41:765–772
40. Hutchings GJ (1991) Effect of promoters and reactant concentration on the selective oxidation of
n-butane to maleic anhydride using vanadium phosphorus oxide catalysts. Appl Catal 72(1):1–32
41. Shekari A, Patience GS (2010) Maleic anhydride yield during cyclic n-butane/oxygen operation.
Catal Today 157(1–4):334–338
42. Pugsley TS, Patience GS, Berruti F, Chaouki J (1992) Modeling the catalytic oxidation of n-butane
to maleic anhydride in a circulating fluidized bed reactor. Ind Eng Chem Res 31:2652–2660
43. Dong Y, Keil FJ, Korup O, Rosowski F, Horn R (2016) Effect of the catalyst pore structure on
fixed-bed reactor performance of partial oxidation of n-butane: a simulation study. Chem Eng Sci
142:299–390
44. Shekari A, Patience GS (2013) Transient kinetics of n-butane partial oxidation at elevated pressure.
Can J Chem Eng 91:291–301

13
1054 Reaction Kinetics, Mechanisms and Catalysis (2019) 126:1027–1054

45. Fernández JR, Vega A, Díez FV (2010) Partial oxidation of n-butane to maleic anhydride over VPO
in a simulated circulating fluidized bed reactor. Appl Catal A 376:76–82
46. Hongbing HZJ, Lefu XHW (2000) Oxidation of n-butane to maleic anhydride over an inorganic
membrane reactor. J Nat Gas Chem 9(3):223–230
47. Emig G, May A, Scheidel P (2002) Computer simulation of the performance of a downer-regenera-
tor CFB for the partial oxidation of n-butane to maleic anhydride. Chem Eng Technol 25(6):627–637
48. Hofmann S, Turek T (2017) Process intensification of n-butane oxidation to maleic anhydride in a
millistructured reactor. Chem Eng Tech 40:2008–2015
49. Ali E, Ali MAH, Alhumaizi K, Elharbawi M (2017) Optimal oxygen feeding policy to maximize the
production of Maleic anhydride in unsteady state fixed bed catalytic reactors. J King Saud Univ Eng
Sci 29:204–211
50. Contractor RM, Bergna HE, Horowitz HS, Blackstone CM, Chowdhry U, Sleight AW (1988) Butane
oxidation to maleic anhydride in a recirculating solids reactor. Stud Surf Sci Catal 38:645–654
51. Gascón J, Valenciano R, Téllez C, Herguido J, Menéndez M (2006) A generalized kinetic model for
the partial oxidation of n-butane to maleic anhydride under aerobic and anaerobic conditions. Chem
Eng Sci 61:6385–6394

Publisher’s Note  Springer Nature remains neutral with regard to jurisdictional claims in published
maps and institutional affiliations.

13

You might also like