A classification model for rock typing using dielectric permittivity and petrophysical data

Dielectric and petrophysical data for both carbonate and sandstone brine-saturated rocks have been used in a discriminant analysis for the purpose of developing a model for rock-type classification. A total of eight petrophysical and dielectric parameters were used in this study. The petrophysical parameters consist of a cation exchange capacity (CEC), a specific surface area (SA) and a rock porosity (φ). The dielectric parameters deduced from the impedance measurements consist of ζs and ζ∞, which are real numbers representing the static and the high-frequency relative dielectric permittivities of the water-saturated rock, respectively, the characteristic relaxation time τ, the spread parameter α and σs, which is the dc conductivity of the water-saturated rock. Outliers have been identified by computing the squared Mahalanobis distances to centroid. Multivariate data cases with relatively large values of the squared Mahalanobis distance associated with small probabilities in the order of 0.001 or less have been removed. Results of the discriminant analysis indicate that only four variables (ζs, ζ∞, φ, CEC) are sufficient to identify rock types. The analysis reveals the existence of a significant discriminant function to distinguish among two distinct rock types related to two broadly defined lithofacies: sandstones and carbonates. A rock-type classification model based on dielectric permittivity and petrophysical data is, therefore, introduced. The model has been validated by an independent set of testing samples. The results of this study indicate that the use of dielectric permittivity data, in conjunction with basic rock properties such as the porosity and the cation exchange capacity, appears to be a robust approach for hydrocarbon rock-type classification