Thèse soutenue

Analyse de l'effet électrostatique et recherche sur la technologie de détection et de localisation des fuites dans le processus de transport des pipelines de fluides pétrochimiques

FR  |  
Auteur / Autrice : Xu Diao
Direction : Ahmed MébarkiJuncheng Jiang
Type : Thèse de doctorat
Discipline(s) : Sciences de l'ingénieur
Date : Soutenance le 05/06/2022
Etablissement(s) : Université Gustave Eiffel en cotutelle avec Nanjing University (Chine)
Ecole(s) doctorale(s) : École doctorale Sciences, Ingénierie et Environnement (Champs-sur-Marne, Seine-et-Marne ; 2015-....)
Partenaire(s) de recherche : Laboratoire : Laboratoire Modélisation et simulation multi échelle (Marne-la-Vallée) - Laboratoire Modélisation et simulation multi échelle (Marne-la-Vallée)
Jury : Président / Présidente : Remming Pan
Examinateurs / Examinatrices : Ahmed Mébarki, Juncheng Jiang, Myriam Mokhtari Merad, Zhou Ru, Xuhai Pan, Yong Pan, Jinghong Wang
Rapporteurs / Rapporteuses : Myriam Mokhtari Merad, Zhou Ru


FR  |  

Pipeline transportation is one of the most common modes of petrochemical fluid transportation. If a pipeline leaks, it may cause serious consequences, such as economic loss, environmental pollution, or fire and explosion. In order to deal with the problems caused by pipeline leakage, this research first aims at investigating the flow electrification of the intact pipe and the leak pipe. The theoretical models able to calculate the space charge density of intact and leak pipes are proposed based on the charge conservation equation. The distributions of the space charge density and electrostatic potential are investigated through numerical simulation. The effects of flow velocity and pipe diameter on the flow electrification of intact and leak pipes are then investigated. Furthermore, the streaming current inside the pipe and the leakage current from the pipe wall to the ground are also studied by experiments. The results show that the streaming current and the leakage current do not change with time. However, the leakage current increases abnormally at the moment the pipe leaks, which may increase the electrostatic hazard.Afterward, the risk of the domino effect of fire or explosion accident caused by petrochemical fluid pipeline leakage is studied based on the fault tree analysis (FTA). 480 minimal cut sets (MCS) and 10 minimal path sets (MPS) are obtained. Qualitative analysis of structural importance is conducted and the condition of formation of the combustible mixture, that is the ‘Continuous leakage’ and ‘Very low wind speed’, has the largest structural importance. Then, the fuzzy failure rate (FFR) of each basic event is calculated by using the quantitative analysis method combining the analytic hierarchy process (AHP) and fuzzy theory, and the occurrence probability of the top event is obtained as 1.3258×10−5. In order to avoid the disastrous consequences caused by pipeline leakage, pipeline leak detection and location method based on transient flow is developed. The leak location model is based on the time when a transient wave propagates from the end valve to the leak location and back again. Furthermore, in the leak detection model, the one-dimensional unsteady friction model is introduced into the method of characteristics (MOC). Then, the governing equations are derived as a ternary system of equations, in which the unknown parameters, especially leak size coefficient, are obtained by analyzing the first transient pressure wave. Both one leak and two leaks situations are investigated and the results show good applicability.Furthermore, a pipeline leak detection and location method based on the acoustic emission (AE) technology is also proposed. An improved signal decomposition method based on particle swarm optimized variational modal decomposition (PSO-VMD) is proposed to denoise AE signals. The 26 time-frequency features of the denoised AE signals are extracted as the input vector for leak detection. The leakage detection methods based on supervised learning (SVM) and semi-supervised learning (PCA) are performed, respectively. The results show that both methods can better identify the leakage condition. The wave velocity for leak location is obtained according to the dispersion guided wave curve of actual pipeline. Finally, the leakage location is carried out according to the time delay estimation, and the location error is between 0.8% ~ 12.1%