Development of a New Correlation for Predicting Initial Water Saturation in Carbonate Reservoirs

Abstract

The Middle East, rich in oil and gas within carbonate rocks, accounts for a significant portion of global reserves, drawing extensive exploration by major oil firms. Unlike Southeast Asia's fracture and cavity-dominated carbonate reservoirs, the Middle East features thick-bedded, pore-structured reservoirs with vast reserves. These complex and varied pore structures cause reservoir inhomogeneity, challenging the technical evaluation of these unconventional reservoirs. Characterization of carbonate reservoirs differs in terms of their mineralogical compositions and heterogenous pore systems from that of clastic reservoirs. Reservoir characterization seeks to build geological and petrophysical models for reservoir simulation. Rock types represent the most crucial characteristics of reservoirs for specialized facies modelling within specific ranges of porosity and permeability. Rock typing is an essential method routinely used by petroleum engineers for characterizing and predicting the reservoir quality of carbonate reservoirs by classifying reservoir rocks into distinct units based on similar petrophysical properties. It is imperative to predict these reservoir properties accurately and precisely. The J-function technique is considered the most effective rock typing procedure. In this study, a new correlation for predicting initial water saturation (Swi) for a reservoir producing from a Permian carbonate formation, located in the Arabian Peninsula, has been developed. The new empirical equation is an augmented Lucia model that utilizes capillary pressure (P�), porosity (), and permeability (k), as independent variables. The coefficient of multiple R2 , the student’s t and F-tests p-value were used in the model evaluation. R2 for the new model was about 0.92, t-test and F-test p-values were much lower than 0.05, indicating that the independent variables are significant. The model was also tested against an independent data set and yielded an R2 of 0.88. Likewise, the new correlation was compared to Lucia’s model and showed better results. The goal of the study is to use the developed correlation in the geostatistical modeling of connate water saturation for analogous formations in the region.