The problem of estimating the effective grid block permeabilities of a fiel
d-scale porous medium with long-range correlations is studied. Both isotrop
ic and anisotropic porous media are considered. The grid blocks are represe
nted by networks of bonds the permeabilities of which are distributed accor
ding to three different stochastic functions that generate long-range corre
lations, two of which are fractal distribution. A new perturbation expansio
n for estimating the effective permeabilities of the system is presented wh
ich, at the lowest order, yields an anisotropic effective-medium approximat
ion (AEMA). The effective permeabilities are also estimated by a renormaliz
ation group (RG) method, as well as computer simulations. The RG method and
AEMA both provide reasonable estimates of the effective permeabilities. Ho
wever, if the porous medium contains zones of very low permeabilities, then
the predictions of the two methods are not very accurate. Two methods are
suggested to increase the accuracy of the predictions. We also show that as
the volume fraction p of the low-permeability zones of the porous medium i
ncreases, the anisotropy of the medium decreases. (C) 2000 Published by Els
evier Science Ltd. All rights reserved.