ARTICLES

Modeling of maleic anhydride production from a mixture of n-butane and butenes in fluidized bed reactor

O.M. Tandioy, I.D. Gil and O.F. Sanchez

Chemical Engineering Department, Universidad de los Andes, Bogotá D.C., Colombia.
o-tandio@uniandes.edu.co, igil@uniandes.edu.co and osanchez@uniandes.edu.co

Abstract - The aim of this work was to model the catalytic oxidation of n-butane and butenes (Raffinate II) mixture to maleic anhydride (MAN) over a vanadium-phosphorus oxide (VPO) catalyst in a fluidized bed reactor (FBR). The three phase Kunni-Levenspiel hydrodynamic model (K-L) was used to describe the reactor. The obtained differential equations were solved by the fourth order Runge-Kutta numeric method. The K-L model was validated by a fitting comparison with reported experimental data for MAN production from n-butane. For the n-butane and Raffinate mixture, the maximum calculated MAN yield was about 52% over the FBR emulsion with a 49% of Raffinate conversion, and 51% of MAN selectivity. As a conclusion, the simulation program demonstrated a suitable performance to predict MAN selectivity, reactants conversion, and MAN and reactants concentration profiles.

Keywords -Maleic Anhydride. Fluidized Bed Reactor. N-butane. Raffinate II.

I. INTRODUCTION

Maleic anhydride (MAN) is the cis-butenedioic acid (maleic acid) anhydride, and is one of the intermediate products with the highest expected demand for the next four years (Table 1), (Brandstädter and Kraushaar-Czarnetzki, 2005; Cortelli, 2006; Gascón et al., 2005). MAN world demand primarily depends on unsaturated polyester resins (UPR) production, lube oil adhesives synthesis and maleic and fumaric acids formation (Nexant ChemSystems Reports, 2005).

Table 1. World MAN production capacity (Cortelli, 2006)

In 1930, National Aniline and Chemical first produ-ced MAN industrially by Benzene catalytic oxidation in a fixed bed reactor (McKetta, 1983). Subsequently, production of MAN has been experimentally investigated by using different reactor types, such as fixed bed reactor (PBR), fluidized bed reactor (FBR), circulating fluidized bed reactor (CFB), two - zone fluidized bed reactor (TZFBR) and membrane reactor. These reactors were developed to improve reaction performance, increase heat and mass transfer rates and adjust reaction features. (Cortelli, 2006; Gascón et al., 2005; Cruz et al., 2005).

Fluidized bed reactors (Fig. 1) have been used in MAN production because of several advantages presented in transport phenomena when catalytic solid particles are suspended and transformed into a fluid-like state (Pugsley et al., 1992).

Figure 1. Fluidized Bed Reactor - Pilot Scale (Lorences et al., 2003).

MAN production in a FBR presents desirable features as follows: (a) prevention of localized hot spot formation via rapid mixing of the catalyst particles, since a constant temperature is maintained throughout the bed by using a simple heat exchanger, even though the reactant and air are fed separately; (b) use of fine particles that allow an increment in the particle contact surface and reduction in the concentration and tempera-ture gradients; (c) catalyst life is increased when hot spots are avoided; (d) feed composition can be 4% butane in air compare to 1.8% in a PBR; (e) the reactor diameter and the compressor investment and cost utili-ties could be reduced since the air rate decreases; (f) easier loading and unloading of the catalyst particles in a FBR than in a PBR; and (g) FBRs need a lower investment than PBRs, since FBRs can have double capacity of a PBR for a large scale plant (Contractor and Sleight, 1987).

Although FBRs have allowed to overcome some maleic anhydride production problems, they still have some drawbacks, such as (a) higher catalyst volume demanded compare to PBR; (b) tendency to lower MAN selectivity; (c) reactants conversion can be affected by a wrong air superficial velocity and an internal reactor design; and (d) it is difficult to keep an oxidative atmosphere at outlet and a fixed conversion range to maintain the MAN selectivity (Yang, 2003).

Initially, maleic anhydride was industrially produced from benzene selective oxidation; this component started to be replaced by n-butane in 1974 because of its toxic effects and economic aspects (McKetta, 1983). In 2003, seventy percent of MAN was produced from n-butane, with MAN selectivities from 53% to 65% (for a PBR or FBR) and n-butane conversions lower than 86% (Cortelli, 2006). Also, MAN has been produced from butenes as raw material, with a 1-butene selective oxidation reaching a MAN selectivity of 48% (Ozkan and Schrader, 1986). In other hand, Raffinate II, a liquefied hydrocarbon mixture (Table 2), has recently been studied for MAN production (Unipetrol Rafinerie, 2007). This substance is coproduced upon the ethylene and propylene manufacture in steamcracker plants (Brandstädter and Kraushaar-Czarnetzki, 2005).

Table 2. Raffinate II composition. (Brandstädter and Kraushaar-Czarnetzki, 2005)

Currently, industry has explored MAN production from partial oxidation of butenes and n-butane mixtures (Brandstädter and Kraushaar-Czarnetzki, 2005). Never-theless catalytic oxidative reactions demand a high operational temperature (higher than 350°C), which means an expensive experimental investment. Thus, process modeling becomes an important and efficient tool to evaluate its performance. In this work, maleic anhydride production from a mixture of butenes and n-butane partial oxidation catalyzed with vanadium phosphorus oxide (VPO) in a fluidized bed reactor is modeled and simulated, applying fourth order Runge-Kutta numeric method to solve the system of ordinary differential equations obtained. All simulations were supported by Microsoft Visual Basic 6.3.

II. HYDRODYNAMIC MODEL

The fluidized bed reactor was defined to operate in the bubbling regime of fluidization. Under this regime, air flow rate is higher than the minimum fluidizing flow rate. As a consequence, agitation is turbulent and solids movement is vigorous; thus, bubbles rising are generated inside the reactor (Kunii and Levenspiel, 1991). There are several hydrodynamic models propo-sed, which can be classified as: one fluidization phase model, which does not have enough accuracy to predict fluidized bed behavior; the two-phase approach, where most of their parameters values are figured out by experimental methods (Harriot, 2002); and the three-phase models that account: bubble, cloud-wake and emulsion phases (Missen et al., 1999).

Based on the good correlation between three-phase hydrodynamic models and fluid bed hydrodynamic behavior in FBRs that operate over bubbling regime, the bubbling bed model proposed by Kunii-Levenspiel (K-L) was selected. This model considers a mass transfer among three phases (bubble, cloud-wake and emulsion); while bubbles rise and reactants diffuse through the phases to get in contact with the catalyst particles at the emulsion. The mass transfer flow rate of the reactants and the products influence feedstock conversion; therefore being required to calculate the bubble rising velocity (u_b) and gas mass transfer coefficients for each phase (Eqs. 1 to 5) (Kunii and Levenspiel, 1991).

	(1)
	(2)
	(3)
	(4)
	(5)
	(6)

To compute the mentioned variables is necessary to define other parameters. These are given by Eqs. 7 to 9 (Kunii and Levenspiel, 1991):

	(7)
	(8)
	(9)

The K-L hydrodynamic model assumes that: (a) bubbles are uniformly distributed in the bed and have the same size throughout the bed; (b) gas flow close to rising bubbles follows the Davidson model, where all parameters are related with the bubble size; (c) each bubble develops a circulation of solids in the bed, with upflow behind bubbles and downflow in the rest of the emulsion; and (d) emulsion keeps at minimum fluidi-zing conditions. As a result, the relative gas velocity and solids remain constant (Kunii and Levenspiel, 1991; Yang, 2003).

III. REACTOR MODEL

The applied model to the FBR assumes that: (a) the reaction is catalyzed by catalyst particles that are suspended in the bed; (b) the reactor operates under isothermal conditions, at uniform density, and at steady state; (c) the reactants rise inside bubbles and cloud as a fluidized gas, and they are in convective flow; further-more, there is no net gas flow through the emulsion phase; (d) the bubbles rise through the bed at plug flow; (e) gas exchange happens from bubbles to cloud-waste phase and from cloud-waste phase to emulsion by using K_bc and K_ce gas transfer coefficients, respectively; (f) radial gradients of temperature and concentration can be neglected, because of the violent movement of the small particles (catalyst) (g) axial dispersion can be ignored since Peclet number is large enough to do so; also, the Peclet number value is between N_r (unit number of reactions in series, Eq. 10) and N_m (unit number of mass transfer, Eq. 11), which guarantees a rapid overall reaction and a mass transfer controlled by the convec-tive flux, respectively. Additionally, the values of Peclet number do not even change if axial dispersion is taken into account (Harriot, 2002).

	(10)
	(11)

The conditions assumed for Eq. 10 and 11 were used to establish the mass balance for each phase over a control volume (Fig. 2). As a consequence, a continuity equation is obtained for all components present at each phase (Eqs. 11 to 13).

	(12)
	(13)
	(14)

Figure 2. Schematic Representation of control volume for material balance at the FBR (Missen et al., 1999)

IV. CATALYST

Catalyst particle properties reported by Lorences et al. (2003), who used VPO prepared in an organic medium and encapsulated in a shell of porous silica to avoid attrition of particles and their entrainment, were used in this work.

A. Catalyst Deactivation

Vanadium present in VPO particles is the main oxygen transporter. The catalyst solid surface is transformed by a vanadium reduction process when it reacts with the hydrocarbon in the gas phase, V⁵⁺ - V⁴⁺ (Mostoufi et al., 2001). The decrease of the V⁵⁺:V⁴⁺ ratio makes the MAN selectivity decline (Cortelli, 2006). VPO catalyst reduction is considered a parallel deactivation, where the reactants become products while they deactivate the catalyst particles. The deactivation model was used by Mostoufi et al. (2001), who assumed a first order deactivation rate and a 90% of deactivation for the catalyst (Eqs. 15 to 17) (Bartholomew and Farrauto, 2006; Murzin and Salmi, 2005; Pugsley et al., 1992):

	(15)
	(16)
	(17)

V. METHOD

A. Hydrodynamic model validation

The hydrodynamic model chosen for the fluidized bed reactor simulation was validated at the experimental conditions reported by Lorences et al. (2003) for the MAN production from n-butane in FBR catalyzed by VPO. The used kinetic model was reported by Buchanan and Sundaresan (1986). These expressions have the form of a single-site redox model (Table 4, Fig. 3, Eqs. 18 to 22).

	(18)
	(19)
	(20)

where j=1 or 2

	(21)
	(22)

Table 4. Size and Operating Conditions for the validated FBR (Buchanan and Sundaresan, 1986).

Figure 3. Mechanism reaction of the n-butane partial oxidation (Buchanan and Sundaresan, 1986).

The fourth order Runge-Kutta numeric method was applied to solve the obtained system of ordinary differential equations (12 ODEs) (Burden, 2002). These simulations were supported by Microsoft Visual Basic 6.3.

B. Modeling of MAN production from a mixture of n-butane and butenes in a FBR

The validated hydrodynamic model was used to model the maleic anhydride production from a mixture of n-butane and butenes keeping catalyst properties (Lorences et al., 2003), using Raffinate II composition and operational conditions reported for MAN production from a mixture of n-butane and butenes in a FBR (Brandstädter and Kraushaar-Czarnetzki, 2005).

The applied kinetic model was reported by Brandstädter and Kraushaar-Czarnetzki (2005) for the Raffinate II oxidation to MAN (Table 5, Fig. 4, Eqs. 23 to 33).

Table 5. Size and Operating Conditions for the FBR modeling (Brandstädter and Kraushaar-Czarnetzki, 2005).

Figure 4. Reaction network for oxidation of mixtures of n-butane and butenes (Brandstädter and Kraushaar-Czarnetzki, 2005)

Kinetics were defined by the following equations, where j=1,2,...,8:

	(23)
	(24)
	(25)
	(26)
	(27)
	(28)
	(29)
	(30)
	(31)
	(32)
	(33)

The fourth order Runge- Kutta numeric method was applied to solve the obtained system of ordinary differential equations (18 ODEs) (Burden, 2002). These simulations were supported by Microsoft Visual Basic 6.3.

VI. RESULTS AND DISSCUSION

A. Hydrodynamic model validation

The n-butane conversion (Eq. 34) and maleic anhydride selectivity (Eq. 35) were computed to assess the hydrodynamic model performance compare to the experimental results for the feeding composition and operating conditions reported by Lorences et al. (2003). These parameters were defined for each phase as (Kunii and Levenspiel, 1991):

	(34)
	(35)

Figure 5 presents the n-butane conversion and maleic anhydride selectivity for the predicted and experimental results with and without catalyst deactivation. The simulated model has a suitable performance of conversion (Fig. 5a), while the simulated selectivity is scattered over the whole range (Fig. 5b). Lorences et al. (2003) results showed a comparable performance for n-butane conversion and MAN selectivity by using Bej and Rao kinetics in a pilot scale FBR. Several kinetic models reported in the literature can suitably represent the relationship between operating conditions and n-butane conversion, but not MAN selectivity, which is more sensitive to the assumed kinetic model (Lorences et al., 2003).

Figure 5. Comparison of experimental and predicted conversion of n-butane (a), and maleic anhydride selectivity (b). Model with catalyst deactivation (B) and without deactivation (K).

Therefore, selectivity and high conversions scatte-ring can be a consequence of the kinetics mechanism used. In this case, the kinetics model does not quantify the carbon adsorbed by the catalyst surface, which can extremely reduce the reaction rate (Lorences et al., 2003).

B. Modeling of MAN production from a mixture of n-butane and butenes in a FBR

Modeling the partial oxidation of a n-butane and bute-nes mixture for the MAN production in a FBR were evaluated under the feeding composition and conditions outlined in Tables 2 and 5. Predicted results were compared with reported by Brandstädter and Kraushaar-Czarnetzki (2005) for a PBR. For the evaluation of these results, the concentration of reactants, intermediate products, byproducts and products were normalized to dimensionless concentration according to the amount of the carbon contained by each specie. The used expressions for concentration, conversion and selectivity calculus were:

	(36)
	(37)
	(38)

Figure 6 illustrates rapid butenes consumption over short residence time at all phases (bubble, cloud and emulsion), while a low decrease of the n-butane feeding concentration was observed. Comparison of predicted concentration profiles with PBR experimental results reported by Brandstädter and Kraushaar-Czarnetzki (2005) showed a reaction time shorter in a FBR than in PBR; consequently, butenes reacted completely at 0.2 g s ml^-1 in a PBR, while in a FBR just take 0.03 g s ml^-1 to disappear in the emulsion phase, where most of the reaction happens (Fig. 6c). Accordingly, the residence time is 0.5 g s ml^-1 in a PBR while just 0.16 g s ml^-1 in a FBR.

Figure 6. Predicted Concentration profiles of n-butane (---), butenes (- -) and n-butane plus butenes (-) for the bubble (a), cloud (b) and emulsion (c) phases.

The concentration profiles presented in Figs. 6 and 7 support hydrodynamic model assumptions, which determine that there are not catalyst particles at the bubble and the cloud as much as at the emulsion phase. Therefore, the consumption of reactants and the genera-tion of products are low in the bubble, but grow in the cloud and are large in the emulsion.

Figure 7. Predicted concentration profiles of MAN (- -), CO_x (-) and intermediary products (---) for the bubble (a), cloud (b) and emulsion (c) phases.

Concentration profiles of products are shown in Fig. 7. Intermediary products from the bubble do not completely disappear at the specified period of time. It is not convenient to extend the counted period of time, since MAN is reduced to CO_x in the emulsion. On the other hand, the maximum maleic anhydride yields obtained for each phase bubble, cloud and emulsion were ~19, 45 and 52%, respectively. The highest MAN yield reported by Brandstädter and Kraushaar-Czarnetzki (2005) was ~48% in a PBR.

MAN selectivity is presented in Fig. 8. There was an evident selectivity maximum at cloud and emulsion phases, but there was just a small trend to form a maximum at the bubble. When butenes completely disappeared, the reaction reached these maximums. Hence, there was not a global maximum MAN selectivity since in the bubble butenes were not wholly consumed (Fig. 6b) over the period of time calculated in this work.

Figure 8. Predicted selectivity profiles versus hydrocarbon conversion of MAN (- -), CO (-), CO₂ (- -- -) and intermediary products (---) for the bubble (a), cloud (b) and emulsion (c) phases, and global selectivity in the FBR (d).

MAN selectivity performance against Raffinate II conversion in the cloud and emulsion phases (Fig. 8), are comparable to the MAN selectivity behavior for a PBR reported by Brandstädter and Kraushaar-Czarnetzki (2005). However, MAN selectivity at the bubble phase was even smaller than CO_x selectivity, whereby global reaction performance decreased in the FBR. The maximum MAN selectivity predicted was about 51% with a 49% of Raffinate II conversion, while Brandstädter and Kraushaar-Czarnetzki (2005) reported a maximum MAN selectivity of 51% with a 90% of Raffinate II conversion in a PBR.

Maleic anhydride selectivity for four different reactants (n-butane, butenes, Raffinate II and a hydrocarbon mixture containing 25% mol per mol of butenes) is shown in Fig. 9. The feeding composition was 4% in air under the same operating conditions (410 °C and 1.3 bar) and kinetics. From these profiles is observed that n-butane had the best performance. Thus, maleic anhydride performance depends on feedstock kind. Actually, it depends on the availability of n-butane in the feedstock, as reported by Brandstädter and Kraushaar-Czarnetzki (2005).

Figure 9. Feestock hydrocarbon composition effect on MAN selectivity ( at 410°C and 1.3 bar). Feeding with n-butane (-), butenes (- -- -), Raffinate II (- -) and a mixture 25% butene mol per n-butane mol (--­).

Selectivity for several individual reaction tempera-tures is plotted in Fig. 10, where is evident that the temperature effect over the global reaction performance can be ignored in the evaluated range (from 400 to 450°C). Similar results were reported by Brandstädter and Kraushaar-Czarnetzki (2005).

Figure 10. Reaction temperature effect on MAN selectivity (400 (Δ), 420 (+) and 450°C (- -))

VII. CONCLUSIONS

The three phases Kunii-Levenspiel model presented a good correlation with the actual hydrodynamic phenomenon in a FBR, although high conversions and most MAN selectivity values are scattered when were compared to those predicted with experimental results.

The maximum yield of MAN is not sensitive to temperature changes and depends on the feed composition. Thus, the maximum predicted value was about 52% in a FBR, being larger than in a PBR (~48%). The reaction performance in a FBR is affected by the low selectivity values in the bubble phase. The maximum MAN selectivity leads to ~51% over ~49% of Raffinate II conversion, but it decreases when butenes disappear. MAN production from Raffinate II oxidation in a FBR has a residence time lower than in a PBR, which means, for this case, that a fixed bed zone reaction has a height of 1.5 m while the fluidized bed zone reaction just requires a height of 0.12 m keeping constant the diameter.

Finally, the simulation program showed a satisfac-tory performance for the prediction of selectivity, conversion and concentration profiles for the MAN production from Raffinate II. Nevertheless, further analysis with more detail considerations should be made to predict correctly the MAN production from a hydrocarbon mixture.

NOMENCLATURE

a	Active catalyst fraction
bbta	Inhibition constant related to n-butane, bar
bbte	Inhibition constant related to butene, bar
Ci,k	Concentration of species i at the phase k.
Dp	Catalyst particle Diameter, µm
D	Reactor Diameter, m.
Dak	Gas diffusion coefficient, m2.s-1.
Ej	Activation Energy, J.mol-1.
IP	Intermediate products (Furan y 1,3 Butadiene)
k	Reaction rate constant
Kab	Gas interchange coefficient, s-1
kd	Deactivation rate constant, s-1
Lf	Reactor Height, m.
n	Molar flow rate, mol/s
Pe	Peclet number
Pref	Reference pressure, 0,26 bar
Pbta	Partial pressure of n-butane, bar
Pbte	Partial pressure of butenes, bar
Q	Volumetric flow rate of gas, m3/min
Ri	Global reaction rate of species i
rj	Reaction rate.
S	Selectivity
t	Reaction time, s
T	Temperature, ° C
uo	Gas superficial velocity, m/s
ums	Minimum velocity of the catalyst, m/s
ubr	Rise velocity of a bubble, m/s
X	Conversion
Y	Normalized Concentration
Greek Symbols
?k	Volume of solids dispersed in k phase, divided by the volume of the bubbles
ρ	Density, kg/m3
ek	Voidage of the phase k
?ci	Number of atoms of carbon in species i.
F	Catalyst sphericity
Subscripts
A	n-butane
b	Bubble phase
c	Cloud - waste phase
e	Emulsion phase
HC	Hydrocarbons
i	Species i
mf	Minimum fluidization
P	Maleic Anhydride
O	Oxygen
s	Catalyst particles
0	Inlet

REFERENCES

1. Bartholomew, C.H. and R.J. Farrauto, Fundamentals of industrial catalytic processes," Wiley (2006).         [ Links ]

2. Brandstädter, W.M. and B. Kraushaar-Czarnetzki, "Maleic anhydride from mixtures of n-butenes and n-butane: effective reaction kinetics," Ind. Eng. Chem. Res. 44, 5550-5559 (2005).         [ Links ]

3. Buchanan, J.S. and S. Sundaresan, "Kinetics and redox properties of vanadium phosphate catalysts for butane oxidation," Appl. catal., 26, 211-226 (1986).         [ Links ]

4. Burden, R., Numerical analysis, Thomson Learning, USA (2002).         [ Links ]

5. Contractor, R.M. and A.W. Sleight, "Maleic anhydride from C-4 feedstocks using fluidized bed reactors," Catal. Today, 1, 587-607 (1987)        [ Links ]

6. Cortelli, C., Heterogeneous Catalysts for gas-phase, Selective Oxidation Reactions, Doctoral thesis, Facoltá di Chimica Industriale, Universitá degli Studi di Bologna (2006).         [ Links ]

7. Cruz-Lopez A., N. Guilhaume, S. Miachon and J.A. Dalmon, "Selective oxidation of butane to maleic anhydride in a catalytic membrane reactor adapted to rich butane feed," Catal. Today, 107-108, 949-956 (2005).         [ Links ]

8. Gascon, J., C. Tellez, J. Herguido and M. Menendez, "Fluidized bed reactors with two-zones for maleic anhydride production: different configurations and effect of scale" Ind. Eng. Chem. Res., 44, 8945-8951 (2005)        [ Links ]

9. Harriot, P., Chemical reactor design, Marcel Dekker Incorporated, USA (2002).         [ Links ]

10. Kunii, D. and O. Levenspiel, Fluidization engineering, Butterworth-Heinemann, Boston (1991)        [ Links ]

11. Levenspiel, O., "Chemical reaction engineering," John Wiley & Sons, USA (1999)        [ Links ]

12. Lorences, M.J., G.S. Patience, F.V. Diéz and J. Coca, "Butane oxidation to maleic anhydride: kinetic modeling and byproducts," Ind. Eng. Chem. Res., 42, 6730-6742 (2003).         [ Links ]

13. McKetta, J., Encyclopedia of chemical processing and design, Marcel Dekker, USA (1983)        [ Links ]

14. Missen, R., C. Mims and B. Saville, Introduction to chemical reaction engineering and kinetics, John Wiley & Sons, USA (1999)        [ Links ]

15. Mostoufi, N., H. Cui and J. Chaouki, "A comparison of two-and single-phase models for fluidized-bed reactors," Ind. End. Chem. Res., 40, 5526-5532 (2001)        [ Links ]

16. Murzin, D. and T. Salmi, Catalytic kinetics, Elsevier Science and Technology Books, USA (2005)        [ Links ]

17. Nexant ChemSystems Reports, Maleic anhydride process evaluation (2005).         [ Links ]

18. Ozkan, U. and G.L. Schrader, "Synthesis, character-rization and catalytic behaviour of cobalt molybdates for 1-butene oxidation to maleic anhydride," Applied Catalysis, 23, 327-338 (1986).         [ Links ]

19. Pugsley T.S., G.S. Patience, F. Berruti and J. Chaouki, "Modeling the Catalytic Oxidation of n-Butane to Maleic Anhydride in a circulating fluidized bed reactor," Ind. Eng. Chem. Res., 31, 2652-2660 (1992)        [ Links ]

20. Unipetrol Rafinerie, Raffinate II Characteristics. http://www.unipetrolrafinerie.cz/html (2007).         [ Links ]

21. Yang , W-C., Handbook of fluidization and fluid-particle systems, Marcel Dekker, USA (2003).         [ Links ]

Received: December 10, 2007.
Accepted: April 10, 2008.
Recommended by Subject Editor José Pinto.