Abstract

Interwell heterogeneous reservoir properties for a California heavy oil field were obtained using geostatistical techniques. Fine-grid, cross-sectional and three-dimensional steamflood simulations were then conducted to evaluate how small-scale heterogeneities and methods to represent them affect predicted performance (steam breakthrough time and oil recovery). Actual log-derived properties (which were first calibrated against core measurements) were used.

Results show that oil recoveries during intermediate times (0.6 to 1.2 pore volume cold water equivalent steam injection) differ by 15-25% when small-scale heterogeneities are included (using kriged or conditional simulation) compared with when uniform properties are used. The difference in predicted recoveries between the kriged and conditional simulations were smaller (generally less than 10%). Inclusion of small-scale heterogeneities resulted in higher recovery and delayed steam breakthrough compared to the uniform permeability case when permeability variations were uncorrelated between the injector and producer. On the other hand, it resulted in lower recovery and earlier breakthrough when permeability variations were large and correlatable. Three dimensional (3-D) heterogeneous flow simulation results show that oil recoveries using kriged geological models are generally higher than those from conditional models. Predicted recovery using the 3-D model has a shape similar to and compares reasonably with field data.