APPLICATION OF IMAGE LOGS FOR ENHANCED RESISTIVITY-BASED WATER SATURATION ASSESSMENT IN ORGANIC-RICH MUDROCKS

Abstract

Organic-rich mudrocks are complex in terms of rock fabric (i.e., the spatial distribution of rock components), which impacts electrical resistivity measurements and, therefore, estimates of hydrocarbon reserves. Conventional resistivity-saturation-porosity methods for assessment of water/hydrocarbon saturation do not reliably incorporate the spatial distribution of rock components and pores in the assessment of fluid saturation. Extensive calibration efforts are required for indirectly projecting the impact of rock fabric on resistivity models. For instance, none of the existing shaly-sand models incorporate a realistic distribution of clay network. This might be acceptable in conventional reservoirs. However, oversimplifying assumptions can cause significant uncertainty in reserves evaluation in organic-rich mudrocks. It should be noted that even the methods which incorporate the realistic distribution of rock components are difficult to calibrate. To address the aforementioned challenge, we introduce a joint interpretation of conventional resistivity and resistivity image logs to improve water saturation assessment by honoring the type of rock component, the spatial distribution of the conductive and non-conductive rock components, and the volumetric concentration of fluids and minerals in the rock. Borehole image logs are a source of high-resolution continuous rock sequence records and can provide detailed rock-fabric-related features. In this paper, we propose a method for the estimation of lamination density and mean resistivity value from image logs within each rock type. These fabric-related features are used to quantify the geometric model parameters for each conductive component of the rock. We use these geometric model parameters as inputs to a new resistivity model that considers volumetric concentration and spatial distribution of rock components for a depth-by-depth assessment of water saturation. The other inputs to the workflow are the volumetric concentration of conductive and non-conductive rock components, electrical conductivity of rock components, and porosity estimates from the joint interpretation of well logs. We successfully applied the proposed workflow to a dataset from the Wolfcamp formation in the Permian Basin in which resistivity image logs were available. We observed a measurable variation in estimated image-log-based geometric model parameters, which were in agreement with the visual content of the images. Incorporation of the estimated rock-class-based geometric model parameters in the resistivity model improved water saturation assessment. Results demonstrated a relative improvement in water saturation estimates of 44.2% and 59.1% against Waxman-Smits and Archie's models, respectively. We then used the estimated geometric model parameters for each rock type for a depth-by-depth assessment of water saturation in one additional well without image logs. This led to a faster and more reliable assessment of water saturation within a certain distance from the well with image logs, where the rock types remain comparable. This distance can be evaluated using variogram analysis. We demonstrated that using the estimated geometric model parameters could improve estimates of hydrocarbon reserves in the Permian Basin by approximately 34%. It should be noted that the proposed method for assessment of geometric model parameters is completely based on the actual spatial distribution of rock components and does not require core-based calibration efforts.