Leak diagnosis for pipelines with multiple branches based on model similarity. (March 2021)

• Record Type: Journal Article
• Authors: Torres, Lizeth; Verde, Cristina; Molina, Lázaro
• Is Part Of: Journal of process control. Volume 99(2021)
• Page Start: 41
• Page End: 53
• Publication Date: 2021-03
• Subjects: Leak diagnosis by stages -- Branched pipeline systems -- Fluid process -- Model similarity; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2020.12.003
• Languages: English
• ISSNs: 0959-1524

Abstract

Abstract: This article presents a procedure, which is formulated in three stages, to simplify the problem of localizing a single leak in a pipeline with multiple branches. The first two stages involve the fulfillment of generic conditions deduced from head loss and flow rate balances for detecting and locating the specific section where the leak is present. The fulfillment of these conditions allows finding the analytical expression for the leak localization, which is the goal of the third stage. The key point of the procedure is the search for the section where the leak is found, which is based on a search for similarity between two types of models that depend, respectively, on measured and estimated hydraulic variables. To support the generality of the analytical expression for the leak localization, it is used to derive previously reported particular expressions for localizing leaks in single-branch pipes. The simplicity of the procedure is demonstrated by numerical simulations. Its performance under real conditions is demonstrated with field data of a 60-km long LPG pipeline in which the correct section with abnormal events is determined, despite uncertain data. Highlights: An approach for detecting and localizing leaks in pipelines with multiple branches. A leak diagnosis approach based on head loss and flow rate balances. A generic analytical solution for the localization of a single leak in pipelines with multiple branches. Leak pre-localization is addressed as a problem of searching for similarity between models.