The proposed methodology utilizes the concept of facies as the basis f
or reservoir characterization. Facies represent the elementary units o
f the reservoir. They are objectively determined through a log-based c
luster analysis, and are fully characterized with petrophysical proper
ties from cores. A stochastic model is used to distribute facies and r
elated petrophysical parameters (porosity, permeability and irreducibl
e water saturation) within the reservoir volume. Several equally proba
ble reservoir realizations are generated from the same model and teste
d in a flow simulator. The realization providing the best history matc
h is selected for production forecasts.