EFFECT OF INTRAPARTICLE CONVECTION ON THE TRANSIENT-BEHAVIOR OF FIXED-BED REACTORS - FINITE-DIFFERENCES AND COLLOCATION METHODS FOR SOLVINGUNIDIMENSIONAL MODELS

The effect of intraparticle convective flow inside large-pore catalyst s (e.g. selective oxidation catalysts) on the transient behavior of fi xed-bed catalytic reactors is analysed by using three different transi ent reactor models: the 1-D, heterogeneous, intraparticle diffusion/co nvection model, the 1-D, heterogeneous, intraparticle diffusion model and the pseudo-homogeneous model as a reference model. Results from th ese 1-D models were compared with those obtained in a previous paper w hen radial dispersion was taken into account. The process start-up ana lysis was achieved by feeding the reactor when heated at the wall temp erature. The high capacity of the catalytic bed leads to a faster prop agation of the transient concentration waves than of the thermal ones, developing sharp concentration fronts during the so-called pseudo-iso thermal behavior of the reactor. When the transient regime inside the solid is considered these shock waves are smeared out showing differen t velocities associated with the intraparticle diffusive and convectiv e transports. Higher reactant conversions are achieved when intraparti cle convection is allowed. Besides being centered on the transient res ponses of large-pore systems, this paper also addresses numerical stud ies to allow a good representation of the dynamic features referred to above. The numerical solution for the model equations was obtained th rough the lines method. The performance of two different methods used on the space variables discretization, orthogonal collocation on finit e elements and finite differences is discussed. A strategy of dividing the bed length into sections solved one after the other to avoid larg e dimension problems was also implemented. Important reductions in com puting times can be obtained by using the pseudo steady-state approxim ation for the intraparticle concentration profile and taking as initia l condition the axial concentration profile corresponding to the pseud o-steady-state isothermal solution. Moreover, accounting for the axial dispersion on the model equations, the numerical integration becomes easier to perform with significative reductions in the CPU times. (C) 1996 Published by Elsevier Science Ltd