Enhancing Electrofacies Prediction With Fuzzy Logic: A Pathway to Improved Subsurface Characterization

Accurate electrofacies prediction is essential for optimizing reservoir management strategies, as it directly impacts well placement and production decisions. Traditional methods like clustering and self- organizing maps often fail to capture finer geological details and struggle with vertical resolution variations in well log data. These limitations can result in an incomplete understanding of the subsurface, leading to inefficient reservoir operations and missed opportunities for enhanced recovery. The objective of this study is to address these challenges by developing a fuzzy logic- based algorithm that accounts for uncertainties in well log data, providing more refined electrofacies classification with minimal input data. This approach aims to improve subsurface characterization across multiple wells and formations using basic well logs, while also incorporating a normalization process to enhance the model's applicability to various logging tools and different well environments. The fuzzy logic-based algorithm was implemented in Python and designed to handle the inherent uncertainties in well log responses. Inputs for the model included Gamma Ray (GR), Density, Neutron, Photoelectric Factor (PEF), and Sonic logs, which were processed using fuzzy membership functions. These functions allowed the model to interpret the data within ranges rather than relying on strict cut-offs, enabling more nuanced classification. To ensure consistency across wells using different logging tools or from different tool providers, a normalization process was applied to the well log data. This step made it possible to standardize the data inputs, increasing the model's flexibility and allowing it to be easily adapted for different logging conditions. The simplicity of the required inputs and normalization ensured that the model remained highly replicable across various formations. Core and rock data were used to validate the electrofacies predictions, ensuring their accuracy and alignment with geological realities observed under microscopic analysis. The application of the fuzzy logic algorithm significantly improved electrofacies prediction, successfully distinguishing ten distinct rock types that aligned with both core sample data and microscopic observations. The model was particularly effective at detecting subtle textural variations in vertical resolution that traditional methods often missed, revealing important geological features critical to reservoir performance. The normalization process enhanced the model's robustness, ensuring consistent results across multiple wells and logging conditions. Despite using a streamlined dataset, the model consistently produced reliable outcomes across wells from different formations. This adaptability and accuracy make the fuzzy logic-based approach an excellent alternative to conventional methods, providing better resolution in electrofacies prediction with minimal data input. The model offers valuable insights for reservoir management, especially in complex or heterogeneous formations, supporting more effective and informed operational decisions.