(32f) Evaluation of Multi-Scale Models for Steam Methane Reforming in a Fixed Bed Reactor

Evaluation
of Multi-scale Models for Steam Methane Reforming in a

Fixed
Bed Reactor

Anthony
G. Dixon

Department
of Chemical Engineering, Worcester Polytechnic Institute,

Worcester,
MA, USA, 01609

Models
of chemical reactions in fixed bed reactors, even at steady-state, must take
into account phenomena occurring over a wide range of scales, from the
catalytic site, to the pellet pore, the catalyst particle and finally to the
reactor tube scale. In reactor tubes of low tube-to-particle diameter ratio (N)
the latter can be broken down further into the tube radius scale and the tube
length scale. Conventional pseudo-homogeneous effective (porous) medium fixed
bed models take into consideration one or both tube scales, while the particle
scale is averaged by use of the effectiveness factor. Heterogeneous effective
medium models which solve explicit equations for diffusion and reaction in the
catalyst pellet also include the pellet scale. Even in these multi-scale
models, the transport parameters must lump several phenomena at various length scales.
The question of whether the models are adequate to predict reactor performance
is usually addressed by comparison to experimental data, an approach which is
compromised by uncertainties in the reaction kinetics and in the correlations
used for the effective transport parameters.

In
this work a different method for the evaluation of two-dimensional
heterogeneous fixed bed reactor models is presented, using 3D CFD resolved
particle simulations to generate a detailed benchmark solution at both the tube
and particle scales. The CFD simulation is run for known kinetics and
fluid/particle properties, which then can be used in the effective medium
models, eliminating the uncertainty about whether the chosen kinetics truly
represent the experimental reacting system. The effective transport parameters
can be obtained from CFD runs in the same bed of particles without reaction,
under conditions which mimic the usual experimental set-up to determine these
parameters, for example flow through a heated-wall tube for the effective
radial thermal conductivity, or mass transfer from coated particles for the
particle-fluid mass transfer coefficient. In this way, transport properties which
represent the fixed bed being modeled are used, instead of introducing a large
degree of uncertainty by choosing one of the many literature correlations.

The methodology described in the preceding paragraphs
is illustrated in this talk for steam methane reforming in a packed bed of 807
spheres at N = 5.96, under conditions typical of the middle of a steam reformer
tube, with specified wall heat flux. The 3D CFD simulations show a detailed
picture of temperature, species and reaction rates at the level of a catalyst
particle or lower (See Fig. 1 for contours of the water-gas shift reaction rate).

From
the CFD simulations, it may be determined what level of detail is necessary for
the 2D + 1D model to represent chosen features of the more detailed 3D model.
The results show that the cross-sectional average profiles require detailed
modeling of the reaction rate distribution (via the radial void fraction) as
well as a radially-distributed axial velocity profile, but not a
radially-varying effective radial thermal conductivity (see Fig. 2 for a
comparison for fluid temperature).

These
models which take account of local scale structure and flow, do not reproduce
the radial profiles well. On the other hand, if a model for the local radial
variation of the effective radial thermal conductivity is used, then more
promising results are obtained (see Fig. 3).