Experimental and Model Based Study of Forced Temperature Profiles and Dosing Effects in a Multi-Stage Membrane Reactor

In the field of chemical reaction engineering intensive research is devoted to the development of new processes, allowing to improve selectivity and yields with respect to valuable intermediate. Typical examples, where alternative concepts are desirable, are partial oxidation reactions or hydrogenations. For such reactions improved performance can be achieved by e.g. optimised stage-wise dosing of one or several reactants (Hamel et al., 2003). The implementation of adjusted dosing profiles can be realised by feeding reactants at certain discrete positions of a fixed-bed reactor or by dosing them in a distributed manner using porous reactor walls, e.g. ceramic membranes (Fig. 1). As a further optimisation parameter not considered up to now is the temperature in the particular stages. This is the focus of the current work. In first part of this contribution results of an experimental investigation are presented. Studied were various stage-wise dosing strategies and/or residence time profiles in combination with the realisation of forced temperature profiles. It is shown that this concept allows significant selectivity and yield improvement of intermediate products in parallel-series reactions (Hamel et al.,2003). As an example reaction, the highly exothermic oxidative dehydrogenation (ODH) of ethane to ethylene on a VOX/γ-Al2O3-catalyst was performed. In a comparative study following concepts were investigated: a) a one-stage, b) a cascade of three packed bed membrane reactors (PBMR) and c) the conventional fixed bed reactor. The experimental set up was described elsewhere (Klose et al., 2003). The operating conditions were chosen to avoid intraparticle transport limitations and gas phase reactions. The ethane concentration was varied therefore between 0.5-1.0 vol. %, the oxygen concentration between 0.35 and 18.3 vol. %. The measurements were conducted in a temperature range between 550 and 650 °C. Based on an earlier theoretical analysis (Thomas, 2003), in the PBMR cascade various oxygen dosing profiles (decreasing, uniform, increasing) and in addition with forced temperature profiles were realised experimentally. In a second part, the formulation, solution and validation of a non-isothermal 2D membrane reactor model will be presented. Profound knowledge is available regarding modelling of conventional tubular catalytic fixed-bed reactors. Radial dependencies of porosity and flow profiles are described typically with simplified 1D models, assuming a fully developed flow (Winterberg et al., 2000). Only a few studies were oriented on 2D calculations or more detailed CFD modelling (Dixon et al., 2003). Obviously the situation is more complex if reactants are also dosed along the reactor axis as in the case of the PBMR. To estimate the related local phenomena a pseudo homogeneous 2D model should be useful. Depending on the membrane structure, diffusion dominated (molecular or Knudsen) and/or convection (viscous-flow) dominated transport have to be considered. A critical issue is therefore the proper formulation of boundary conditions at the reactor wall in the mass and energy balances. In our work mass transfer through the wall was described e.g. by the extended Fick (EFM) or the dusty gas model (DGM). Furthermore special care was taken in order to choose appropriate correlations describing radial mass and heat transport in the packed bed. The main focus of the parametric study performed was the estimation of internal temperature, oxygen concentration and flow profiles and the resulting effects on the reactor performance. As in the experimental study the ODH of ethane was chosen as a model reaction. LHHW reaction kinetics obtained from an early experimental study (Klose et al., 2004) were assumed to describe the reaction network. Selected simulation results obtained for a constant and a radial dependent porosity profile are shown in Fig. 2. The calculations were performed using the commercial software packed Femlab 3.0TM. These results show the relevance of two-dimensional modelling. References Dixon, A.G., M. Nijemeisland, H. Stitt, Int. J. of Chem. Reactor Eng. 2 (2004) A24. Hamel, C., S. Thomas, K. Schädlich, A. Seidel-Morgenstern, Chem. Eng. Sci. 58 (2003) 4483. Klose, F., M. Joshi, C. Hamel, A. Seidel-Morgenstern, Appl. Catal. A. 260 (2004) 101. Klose, F., T. Wolff, S. Thomas, A. Seidel-Morgenstern, Catal. Today 82 (2003) 25. Thomas, S., Ph.D. Thesis, Otto von Guericke University, Magdeburg (2003) Winterberg, M.,E. Tsotsas,A. Krischke, D. Vortmeyer, Chem. Eng. Sci. 55 (2000) 96