Prioritizing Well Logging Based on Corrosion Risk Assessment Using an Advanced Artificial Intelligence Driven Approach

Abstract Downhole casing corrosion presents a persistent threat to the integrity of oil and gas wells. Effective corrosion risk management is crucial to prevent unforeseen equipment failure, compromising well performance leading to catastrophic safety events. This paper presents an innovative proactive (AI)-driven framework for predicting well-specific corrosion risk profiles to assist production engineers to identify logging interventions requirements and optimize allocated resources. The solution applies a sophisticated regression algorithm, incorporating pertinent variables encompassing well age and design, operating conditions, and environmental factors. These inputs collectively contribute to a precise characterization of each well's unique corrosion susceptibility. Following thorough data discovery, transformation, and quality control procedures, our predictive model yielded probabilistic outputs distinguishing high-risk from low-risk wells. Visualization of predicted outcomes via GIS-enabled spatial heat maps empowered engineers and decision-makers to intuitively grasp complex relationships between wells, identify clusters of elevated corrosion risk, and pinpoint candidates wells for expedited well logging evaluation. The work revealed compelling evidence supporting the effectiveness of this AI-driven strategy, demonstrating an impressive prediction accuracy rate exceeding 90% in identifying high-risk wells. The combination of established heat maps and AI-predicted well risks enabled the generation of a corrosion risk prioritization factor for non-logged wells. This allowed for focused attention on high-risk wells, guiding selective logging considerations and enhanced surveillance to ensure their integrity. Application of this methodology allowed for careful allocation of limited resources toward high-priority wells exhibiting corrosion vulnerability, effectively reducing the cost and downtime associated with unexpected failures. Conversely, lower-risk wells benefited from extended maintenance intervals, fostering substantial reductions in operational expenditures. By targeting logging interventions based on predicted corrosion risk, operators can mitigate potential asset failures, promote improved safety standards, and optimize resource allocation. This study presents a pioneering approach to corrosion risk mitigation in upstream operations by integrating machine learning, statistical modeling, in addition to geospatial analytics. Our work enhances the development of holistic digital simulations for casing corrosion, enabling optimized operations and improved field safety.