Permeability predictions are a critical aspect of reservoir description and are particularly challenging for carbonate reservoirs because of the complexities rooted in diagenesis, grain size variations, cementation etc. Conventional multiple regression techniques very often yield unsatisfactory results.
The paper proposes a simple approach to obtain permeability correlations in heterogeneous carbonate reservoirs using commonly available well logs. The approach follows a two-step procedure. First, the well log data is classified into electrofacies types. This classification is based on the unique characteristics of well log measurements reflecting minerals and lithofacies within the logged interval. A combination of principal component analysis and model-based cluster analysis is used to identify and characterize electrofacies types. Second, a non-parametric regression technique has been applied to predict permeability using well logs within each electrofacies. The main advantage of this technique is that it is primarily data driven as opposed to model driven and does not require a priori specification of functional forms, which makes conventional multiple regressions difficult.
This method has been successfully applied to a heterogeneous carbonate reservoir, the Heera and the South Heera field in offshore India. The field is producing for over twenty five years and a major redevelopment effort is ongoing to improve production from the field by targeting the bypassed oil. Geologic modelling and flow simulations are important aspects of the redevelopment planning efforts. Our proposed approach results in improved predictions of permeability using commonly available well logs. The permeability correlations were incorporated into the geologic modelling to generate field-wide 3-D distributions of permeability. A scaling of the cumulative probability distribution was carried out to modify these permeabilities to account for well test data. Preliminary simulation runs indicate that the 3-D geolgic model is able to adequately reproduce the pressure trends and the waterfront movement in the reservoir. Detailed history matching is currently in progress.