Integrating Machine Learning, Improved Geostatistical Modeling and Experimental Design Techniques for Reservoir Characterization and Probabilistic Reserves Estimation: A Niger Delta Case Study

Abstract Accurate reserves estimation and uncertainty quantification are essential for optimizing hydrocarbon recovery and guiding strategic field development. For fields in the Niger Delta where undeveloped reservoirs present opportunities for increased production, the challenge lies in overcoming data sparsity and improving reservoir characterization. This paper evaluates two prolific deep reservoirs in the Niger Delta requiring further appraisal and resource estimation to unlock their potential. To address these challenges, an integrated, multidisciplinary workflow was deployed, using advanced machine learning techniques, improved geostatistical modeling and experimental design techniques to further gain better understanding of the subsurface. A supervised machine learning algorithm leveraging better horizontal resolution of seismic was employed to predict well logs in data-sparse regions, facilitating enhanced petrophysical modeling and facies classification. Reservoir properties were distributed within a 3D static model using geostatistical techniques, while flow connectivity was assessed through geo-screening to ensure viable reservoir communication. Pressure-Volume-Temperature (PVT) and relative permeability data, along with fluid contacts, were integrated for robust model initialization, and probabilistic production performance predictions were conducted using proxy modeling techniques. This comprehensive workflow delivers significant benefits in reducing uncertainties and enhancing decision-making in field development. By integrating advanced data-driven methods with traditional reservoir modeling and reserves estimation techniques, the approach provides actionable intelligence for maximizing hydrocarbon recovery and optimizing field planning. It underscores the practical value of innovative methodologies in addressing complex subsurface challenges and driving strategic outcomes.