The performance of in situ air sparging is controlled by the distribution o
f air pathways in the subsurface, which is in turn controlled by the struct
ure of the medium to be sparged. The specific pathways that the air follows
are determined, at the grain scale, by the distribution of air entry press
ures of the pores. At the field scale, pore size distributions are usually
correlated with heterogeneous structures (e.g. layers) within the medium, w
hich control the macroscopic distribution of the air. The processes that pr
oduce an observed air distribution at a particular site are complicated, an
d are potentially well suited to modeling with multiphase flow models. Rece
nt numerical modeling of heterogeneous media appears to successfully repres
ent expected distributions of air; however, current models do not provide a
tool to predict sparging performance. For this to be the case, the models
need to represent the detailed structure of the medium at the site to be st
udied, as well as to capture the relevant aspects of the discrete air/water
distribution (i.e., represent air channels at the centimeter or smaller sc
ale). This will, in general, require a level of site data that is not avail
able and numerical models that require many millions of computational eleme
nts. As a consequence, at least for the foreseeable future, numerical model
ing of the air sparging process will continue to play a vital role as a con
ceptual tool with Limited predictive capability at sites. (C) 2000 Elsevier
Science B.V. All rights reserved.