Session: 11-07-01 Petroleum Production: Offshore Systems and Subsea Operations

Submission Number: 157014

AI-Based Adaptive Digital Twin Framework for Real-Time Leak Detection and Localization in Offshore Gas Pipelines 

Digital twins are revolutionizing the digitalization and automation of offshore systems, playing a pivotal role in enhancing real-time monitoring, predictive maintenance, and operational efficiency in offshore gas pipeline networks. This paper introduces a novel adaptive digital twin framework for leak detection and localization in offshore gas pipelines, employing advanced machine learning (ML) techniques and artificial intelligence (AI) models that are specifically optimized for adaptability across varying pipeline configurations. The core innovation lies in a calibration process that combines OLGA--derived leak detection monographs—originally tailored to specific pipeline profiles based on pressure drop and mass flow data—and recalibrates them to seamlessly adapt to new pipeline geometries and operating conditions.

Through the use of neural networks and other ML algorithms, including transfer learning and ensemble models, the framework dynamically updates leak detection models with minimal reconfiguration, enabling rapid deployment in diverse pipeline scenarios. This recalibration methodology allows for the generation or adjustment of leak detection monographs for new pipelines, ensuring accurate real-time predictions of leak size and location. Enhanced by digital twin visualization, the framework provides actionable feedback on leak events, bolstering operational safety and minimizing environmental impact by swiftly addressing leak incidents.

The study addresses key challenges in offshore gas pipelines, such as high-pressure effects, complex flow profiles, and variable environmental conditions, which can significantly influence leak characteristics. By demonstrating reliable leak detection and localization with reduced recalibration efforts, our approach offers scalable and practical solutions for offshore flow assurance and pipeline integrity management. This adaptive digital twin framework not only enhances risk mitigation and predictive maintenance but also supports autonomous operational management in offshore oil and gas operations, contributing to industry-wide digital transformation goals. The proposed approach provides a pathway for scalable deployment of AI-driven digital twins in offshore environments, with significant implications for enhancing flow assurance and ensuring the integrity of critical gas pipeline infrastructure.

Presenting Author: Mohammad Azizur Rahman Hamad Bin Khalifa University (HBKU)

Presenting Author Biography: Dr. Mohammad Azizur Rahman received his Ph.D. in Multiphase Flow from the University of Alberta, Canada, in 2010. He is currently an Associate Professor at Hamad Bin Khalifa University (HBKU), where he contributes to advancing research and teaching in the field of petroleum engineering. Prior to joining HBKU, Dr. Rahman was an Associate Professor at Texas A&M University at Qatar, a position he held from 2016. Earlier in his career, he served as an Assistant Professor at Memorial University of Newfoundland and an Instructor at the University of Alberta, Canada.

Dr. Rahman teaches and conducts research in production engineering, specializing in multiphase flow. He has received around $3 million in research funding from organizations such as Qatar Foundation, the Natural Sciences and Engineering Research Council of Canada, and the Newfoundland Research & Development Corporation. He has collaborated with leading companies, including TotalEnergies, SLB, QatarEnergy LNG, Intecsea, NEL, Syncrude Canada, GRI Simulations, C-Core, and Petroleumsoft.

Dr. Rahman is actively involved with professional organizations such as SPE and ASME and is a registered Professional Engineer in Alberta, Canada. He has established a multiphase flow loop and made significant contributions to the field, with more than 250 refereed journal and conference publications focusing on multiphase flow experiments and computational fluid dynamics simulations.