Digital twin for leak detection and fault diagnostics in gas pipelines: A systematic review, model development, and case study

Leak detection in oil and gas pipelines plays a crucial role in ensuring flow assurance. Numerous analytical and experimental methods are employed in the industry for leak detection. However, many of these techniques are costly or suffer from poor performance, including high false alarm rates. This research aims to develop a robust digital twin (DT) to address these issues. A comprehensive review is conducted, focusing on evaluating the performance of existing leak detection methods, including DT-based models. Furthermore, an overview of machine learning techniques used in the development of pipeline DT models is provided. The review reveals that current DT models are primarily focused on leak detection but do not adequately identify leak size and location. This study proposes a more comprehensive DT model capable of detecting pipeline abnormalities, such as leaks, equipment failure, and damage. The proposed DT is applied to a real-field gas pipeline case study to validate its feasibility. The results demonstrate that the developed DT model successfully detects leaks with a zero false alarm rate and accurately identifies leak size and location with an absolute relative error of less than 3.21 %. This work serves as a reference for future DT development and research in pipeline monitoring.