Lithotype is shown to be the main geological control on the spatial distribution of reservoir flow properties. The distribution of lithotypes is in turn controlled by lithofacies. Using core data, assemblages of lithotypes were grouped into major facies associations (MFA's). Log signatures were used to pick MFA's in uncored wells to provide conditioning data for a stochastic description of their interwell distribution using the sequential indicator simulation (SIS) technique. Deterministic correlation was not enforced.
Characteristic object-models of the distribution of lithotypes within each MFA were then generated and converted to fine-scale poro-perm models using core-plug data. These models were ups-caled to yield characteristic poro-perm distributions at the reservoir simulation scale. Finally, the MFA model was used as a template to distribute the upscaled poro-perm within the simulation model. The model history-matched rapidly and accurately, even though the wells were conditioned only to the MFA's and not to foot-by-foot data. Multiple MFA realizations were generated to provide some understanding of the uncertainty in thickness and other rock properties between wells.