Effective dispersion model for flow-through catalytic membrane reactors combining axial dispersion and pore size distribution

Abstract

In a flow-through catalytic membrane reactor, catalyst is immobilized in the pores of a membrane, which is convectively passed by the reaction mixture, allowing for high catalytic activity and a potentially narrow residence time distribution (RTD). As the membrane geometry prevents measurement of a meaningful RTD, a residence time distribution model is introduced, which accounts for deviations from ideal plug flow behavior induced by a non-ideal pore size distribution and by axial molecular diffusion. Both effects are combined to an effective dispersion model with a single dimensionless parameter, which is a function of pore geometry, axial velocity and molecular diffusion coefficient. As a result of the developed model, recommendations for optimum reactor operation are given.

Introduction

The term “flow-through catalytic membrane reactor” describes a reactor concept for heterogeneous reactions, where the catalyst is immobilized in the pores of a mostly ceramic membrane, which are convectively passed by the reaction mixture. The porous membrane does not perform any separative tasks and is solely used as microstructured catalyst support. This type of reactor allows for high catalytic activity due to intensive contact between reactants and catalyst and potentially for a narrow residence time distribution (Zaspalis et al., 1991, Julbe et al., 2001, Sanchez Marcano and Tsotsis, 2002, Dixon, 2003, Dittmeyer et al., 2004, Schomäcker et al., 2005, Westermann and Melin, 2009).

The residence time distribution of the reactants inside the catalytic pores is of major interest, if for example a sequential reaction with a desired intermediate product is performed. In this case the highest selectivity can be achieved in an ideal plug flow reactor. The flat membrane geometry with rather large lateral dimensions in the order of centimeters compared to the extremely short reactor length of a membrane pore prevents the direct measurement of a meaningful residence time distribution, because the time a tracer spends for spreading radially across the membrane without contact to catalyst is at least in the same order of magnitude as the time it spends inside the catalytically active pores.

In order to at least qualitatively assess the hydrodynamic behavior inside the catalytic pores, this work introduces a residence time distribution model for flow-through membrane reactors, which accounts for the main deviations from ideal plug flow. Non-ideal reactor behavior results from maldistribution of the flow due to a pore size distribution (Fig. 1) on the one hand and from axial dispersion inside the pores induced by molecular diffusion on the other hand. The model is based on the straight microchannel structure of anodized membranes. It might be applied for sintered membranes as well, but would require implementation of suitable correction factors such as tortuosity.