ABSTRACT
Abstract Internal and external mass transfer limitations in porous catalyst layers play a central role in three-phase processes. The governing phenomena are well-known since the days of Thiele (1) and Frank-Kamenetskii (2). Transport phenomena coupled to chemical reactions is not frequently used for complex organic systems. A systematic approach to the problem is presented. Industrially relevant consecutive-competitive reaction schemes on metal catalysts were considered: hydrogenation of citral, xylose and lactose. The first case study is relevant for perfumery industry, while the latter ones are used for the production of sweeteners. The catalysts deactivate during the process. The yields of the desired products are steered by mass transfer conditions and the concentration fronts move inside the particles due to catalyst deactivation. The reaction-deactivation-diffusion model was solved and the model was used to predict the behaviours of semi-batch reactors. Depending on the hydrogen concentration level on the catalyst surface, the product distribution can be steered towards isomerization or hydrogenation products. The tool developed in this work can be used for simulation and optimization of stirred tanks in laboratory and industrial scale. Introduction Most parts of heterogeneously catalyzed reactions are carried out in porous catalyst layers, such as catalyst particles in fixed beds, fluidized beds, slurry reactors or in catalyst layers in structured reactors, such as catalytic monoliths, fibers or foams (4 - 6). In the manufacture of organic chemicals, three-phase processes are very frequent, catalytic three-phase hydrogenation being the most typical example. Such manufacturing processes very often involve a strong interaction of kinetics, mass transfer limitations and catalyst deactivation. Mass transfer might play a central role in various stages of a catalytic three-phase process: at the gas-liquid interface as well as inside the porous catalyst layer. In the presence of organic components, catalyst deactivation due to poisoning and fouling retard the activity and suppress the selectivity with time. At elevated reaction temperatures, sintering of the catalyst can take place. Production of organic fine and specialty chemicals is often carried out in
batch and semibatch reactors, which imply that time-dependent, dynamic models are required to obtain a realistic description of the process. Even though the governing phenomena of coupled reaction and mass transfer in porous media are principally known since the days of Thiele (1) and FrankKamenetskii (2), they are still not frequently used in the modeling of complex organic systems, involving sequences of parallel and consecutive reactions. Simple ad hoc methods, such as evaluation of Thiele modulus and Biot number for firstorder reactions are not sufficient for such a network comprising slow and rapid steps with non-linear reaction kinetics. In the current work, we present a comprehensive approach to the problem: dynamic mathematical models for simultaneous reactions media, numerical methodology as well as model verification with experimental data. Design and optimization of industrially operating reactors can be based on this approach. Porous Catalyst Particles Reaction, diffusion and catalyst deactivation in porous particles is considered. A general model for mass transfer and reaction in a porous particle with an arbitrary geometry can be written as follows:
i ρε 1 (1) where the component generation rate (ri) is calculated from the reaction stoichiometry,
jjiji aRr ν where Rj is the initial rate of reaction j and aj is the corresponding activity factor. For the diffusion flux (Ni) various approaches are possible, ranging from the complete Stefan-Maxwell set of equations to the simple law of Fick (7). The symbols of eqs. (1)-(2) are defined in Notation. The catalyst activity factor (aj) is time-dependent. Several models have been proposed in the literature, depending on the origin of catalyst deactivation, i.e. sintering, fouling or poisoning (8). The following differential equation can represent semi-empirically different kinds of separable deactivation functions,
and
dc DN ieii , where Dei=(εp/τp)Dmi according to the random pore
model. A further development of eq. (1) yields
The boundary conditions of eq. (5) are
dci at 0=r (6a) and
where ci is the bulk-phase concentration of the component. The initial condition at t=0 is ci(r)=ci, i.e. equal to the bulk-phase concentration. The molecular diffusion coefficients in liquid phase were estimated from the correlations Wilke and Chang (9) for organic solutions and Hayduk and Minhas (10) for aqueous solutions, respectively. The solvent viscosities needed in the correlations were obtained from the empirical equation based on the experimental data,
(s=2) and trilobic s=0.49, were used in the experiments. Hydrogen solubility in the organic reaction mixture was obtained from correlation of Fogg and Gerrard (12), ( ) TBAxH /ln * 2 += , where is the mole fraction of dissolved hydrogen in equilibrium, and A and B are experimentally determined constants. For aqueous sugar solutions, a correlation equation proposed by Mikkola et al. (13) was used.
