a b s t r a c t The effect of catalyst pore structure on n-butane oxidation to maleic anhydride in a fixed-bed reactor was investigated by numerical simulations. The micro- and macro-pore model of Wakao and Smith was applied to model the diffusion–reaction inside the catalyst pellet. The studied pore structure parameters were macro-pore porosity, mean macro-pore diameter and mean micro- pore diameter. A fixed-bed reactor was simulated with a detailed two-dimensional heterogeneous model under typical industrial conditions. Simulation results have demonstrated that the reactor performance is sensitive to the chosen pore structure parameters especially macro-pore porosity and mean micro-pore diameter. A bi-modal catalyst pellet with bigger macro-pores and smaller micro-pores is favored to achieve higher yields of maleic anhydride. This work highlights the potential of improving this process by pore structure opti- mization.69811

& 2015  Elsevier  Ltd.  All  rights reserved.

1. Introduction

Maleic anhydride (MAN) is an important intermediate in che- mical industry with an annual worldwide consumption of 2.7 Mt (Trifirò and Grasselli, 2014). Since 2006, over 70% of the MAN is produced by selective oxidation of n-butane catalyzed by a vana- dyl pyrophosphate (VPP) catalyst. Processes employ either multi- tubular fixed bed reactors immersed in a molten salt bath or fluidized bed reactors (Schunk, 2008; Guliants et al., 2000). The fixed-bed process, which we will focus on in this work, runs typically at 80–85% n-butane conversion and the non-converted butane is not recycled. The overall yield of MAN reported is 57– 65% which is not optimal for selective oxidation processes   (Trifirò

and Grasselli, 2014). Therefore, there is still  room  for improving the fixed-bed process and maximizing  the yield  of  MAN.

Nowadays, reactor simulation plays a key role in process opti- mizations as it offers a cost-effective approach compared to experimentation (Stitt et al., 2015). For modeling fixed-bed reac- tors, the pseudo-homogeneous and heterogeneous models are the most popular ones due to their high accuracy and low computa- tional cost (Jakobsen, 2014). In those models, the fixed-bed reac- tors are modeled as a porous media and effective parameters for heat and mass transfer have to be used (Tsotsas and Jcesser, 2010). The pseudo-homogeneous and heterogeneous models have been applied to simulate the n-butane oxidation in fixed-bed reactors (Sharma et al., 1991; Ali and Al-Humaizi, 2014; Diedenhoven et al., 2012; Brandstädter and Kraushaar-Czarnetzki, 2007; Guettel and Turek, 2010) and membrane reactors (Marín et al., 2010; Alonso et    al.,    2001).    The    main    difference    between    the pseudo-

homogeneous and heterogeneous model is that the heterogeneous model considers explicitly the presence of the catalyst phase. Therefore, two sets of conservation equations are applied for the interstitial fluid and the catalyst phase, respectively (Jakobsen, 2014). The resolved concentration and temperature profiles inside the catalyst pellets are coupled to the bulk fluid phase gradients

(Wellauer et al., 1986). The formation of other by-products such as acetic acid, acrylic acid, phthalic and methacrylic acids is generally neglected in the reaction model because of observed low con- centrations (lower than 2 %) (Alonso et al., 2001; Marín et al., 2010).

The stoichiometric equations used were as   follows:

via the boundary conditions applied on the pellet surface. There-

fore, the heterogeneous model can be a useful tool for studying the

catalyst scale parameters to the reactor performance.

The most applied VPP catalysts in industrial fixed-bed reactors