Efficient Model Selection Based on Connectivity for Geologically Complex Reservoirs

To understand how hydrocarbon flow is influenced by the geological properties of channelized reservoirs an experimental design approach is employed. We examined the influence of geometrical properties (channel dimensions and sinuosity), petrophysical properties (permeability and net-to-gross), and derived properties (connectivity). It is found that the geometrical properties have a diffuse relationship with reservoir flow behavior, which is characterized by net present value (NPV). A strong correlation is found for net-to-gross and permeability. Connectivity shows the strongest correlation to production data. Calculating a quadratic response surface of connectivity versus NPV allows that an estimate can be made of the connectivity based solely on NPV and production data. Subsequently, based on the estimated connectivity an efficient selection from a large ensemble of reservoir model realizations can be made. It is demonstrated that there is a high similarity between the reservoir flow behavior of the selected reservoir models and the behavior of the synthetic truth. This suggests that connectivity can serve as a screening technique and can be used as a first step in history matching of model realizations which are created by any automated modeling technique.