GeoPixAI: From Pixels to Intelligent, Unbiased and Automated Fast Track Subsurface Characterization

The core analysis is one of the current inefficacies in rock characterization and petrophysical evaluation practices directly impacting well-completion decisions. The second problem is the lack of an intelligent multi-scale image processor essential for integrating the data into digital 4.0 platforms, which has a significant impact on G&G teams’ decisions for better well completion and reservoir evaluation. In this paper, we discuss the example of slab core analysis using our proposed GeoPixAI workflow. In this application, two different Convolution Neural Networks (CNNs) are designed to measure oil saturation, lithology, and fracture distributions. The results are then compared to the lab measure values. Our techniques show great accuracy while producing results at a fraction of the time it takes to do lab analysis. Also, the results are found to be consistent with lab measurements. While this study focuses on slab core analysis, GeoPixAI is a project focus on Digital Subsurface Processing for intelligent segmentation and information extraction from “raw” images at any scale. GeoPixAI can analyze and extract information from core Micro-CT scans, seismic data, FMI logs, and other image data.