AI Tackles Oil Challenges
In response to the unpredictable nature of global energy markets, exacerbated by geopolitical shifts and supply chain disruptions, a team of researchers
at MIT World Peace University (MIT-WPU) in Pune has pioneered sophisticated artificial intelligence (AI) and machine learning (ML) techniques. Their innovative work is aimed at significantly improving the extraction of oil from mature reservoirs and refining the accuracy of production forecasts. These advancements are particularly crucial for the sustainability of existing oil fields, offering a data-driven approach to overcome the complexities inherent in petroleum reservoir management. The Department of Petroleum Engineering at MIT-WPU is at the forefront of this initiative, developing tools that promise to revolutionize how oil is extracted and managed in an era of global energy uncertainty.
Optimizing Recovery Methods
A dedicated team, spearheaded by Dr. Rajib Kumar Sinharay, a Professor in the Petroleum Engineering department, alongside his PhD student Dr. Hrishikesh K Chavan, has engineered a cutting-edge machine learning model. This sophisticated system is designed to identify and recommend the most effective Enhanced Oil Recovery (EOR) techniques for intricate reservoir environments. By analyzing vast datasets compiled from numerous oil-producing sites across the globe, the model demonstrated an impressive 91% accuracy in predicting the optimal recovery strategies. This breakthrough, documented in the esteemed journal Petroleum Science and Technology, drastically slashes the evaluation time for recovery plans, reducing a process that previously took months to mere hours. Dr. Sinharay emphasized the transformative potential of AI in reservoir management, stating that these data-driven tools empower operators to select superior recovery methods and enhance production forecasting precision, especially for mature oil assets.
Classifying Reservoir Rocks
Parallel to the advancements in recovery techniques, Professor Samarth Patwardhan and his PhD candidate, Dr. Soumitra Nande, have developed a highly accurate deep learning model. This model excels at identifying carbonate reservoir rocks, achieving an exceptional accuracy rate of 97%. Carbonate formations are geologically significant and share similarities with rock types found in Bombay High, India's largest offshore oil field. The research, published in the Arabian Journal for Science and Engineering in 2025, highlights the model's capability to precisely classify these crucial geological structures. Such precise identification is vital for understanding reservoir potential and planning extraction efforts effectively, further contributing to the broader objective of enhancing oil recovery through advanced computational intelligence.
