553
MONTE CARLO SIMULATIONS OF EFFECTIVE DIFFUSIVITIES
IN THREE-DIMENSIONAL PORE STRUCTURES
Sebastian C. Reyes*, Enrique Iglesia*, and Yee C. Chiew**
*Corporate Research Laboratories, Exxon Research and Engineering Company,
Annandale, NJ 08801,
**Department of Chemical and, Biochemical Engineering, Rutgers University,
Piscataway, NJ 08855.
ABSTRACT
A hybrid discrete/continuum Monte Carlo technique combining random
walk simulations with first passage time (FPT) concepts is developed here
in order to estimate diffusion properties of randomly-assembled sintered
porous structures. This work combines the creation of realistic porous
solid structures with controlled pore size, shape, and tortuosity features
with the application of an efficient algorithm for calculating effective
diffusivities in all diffusion regimes (Knudsen, transition, and
molecular). The hybrid simulation technique consists of creating a
"protective" boundary layer surrounding solid surfaces within which
discrete random motion simulations are performed while continuum FPT
results are used in the remaining pore space. The boundary layer thickness
reflects a characteristic length scale, of the order of a few mean free
paths, over which the FPT approximation breaks down. This procedure
significantly reduces the computations required to cover statistically
representative regions of the porous structure, a serious shortcoming in
previous studies of gas diffusion through porous solids; it leads to
effective diffusivity estimates that are in excellent agreement with
experimental measurements.
INTRODUCTION
Chemical reaction rates within porous materials are often controlled
by reactant and product diffusion. Heterogeneous catalysis and
non-catalytic gas-solid reactions are two important areas where diffusion
plays a crucial role. Many obstacles currently prevent the a-priori
estimation of effective diffusivities in porous materials. For example,
several important geometrical and topological properties, such as local
fluctuations in pore size, shape, and pore space connectivity, cannot be
directly measured with available characterization techniques. These
limitations have resulted in many simplified structural pore models. They
consist mostly of cylindrical capillaries, randomly interconnected,
unconnected, or obeying some prescribed topological rules [1]. These
models do not rigorously include dead-ends, irregular pore branching, and
fluctuating pore dimensions that are typical of porous materials.
Therefore, in order to describe experimental data, these models require
adjustable parameters, usually in the form of non-rigorous tortuosity
factors, thus defeating the purpose of providing an a-priori estimation
procedure.
The creation of more realistic three-dimensional models of porous
solids together with Monte Carlo simulations of diffusion were pioneered by
Evans and co-workers ([2] and references therein). Advances on surface and
pore volume characterization continue to provide great opportunities for
improving the accuracy of the simulated structures. Moreover, increasing
computational capabilities allow the incorporation of greater levels of
Mat. Res. Soc. Symp. Proc. Vol. 195. ©1990 Materials Research Society