Gas diffusion in random binary media

The binary representation of porous media can provide a direct means for the characterization of their internal structure and is becoming progressively fashionable as a result of the rapid development of novel, powerful image analysis techniques. In the present work wr provide accurate predictions of the Knudsen and intermediate diffusivities based on molecular trajectory computations in three-dimensional pixelized porous media. The main advantage of this approach is that it avoids resorting to the commonly used concepts of parr, grain, or fiber models. which introduce inevitably a significant degree of approximation to the actual structure. a second advantage is that this method is valid even far from the continuum limit where rarefied transport prevails. An analytical expression for the Knudsen diffusivity in random binary media is suggested that combines simplicity and accuracy over the entire porosity range. In the intermediate diffusion regime, it is found that the serial combination of the overall bulk and Knudsen diffusional resistances provides excellent estimates of the effective diffusional resistance and, hence, a Bosanquet type of approximation is quite valid. Finally, numerical results for the accessible porosity and specific surface area in random binary media are presented using large random realizations that, practically, eliminate statistical errors. (C) 1998 American Institute of Physics. [S0021-9606(98)70440-4].