Portal de Periódicos da CAPES

Emergency response recommendation for long-distance oil and gas pipeline based on an accident case representation model

2022; Elsevier BV; Volume: 77; Linguagem: Inglês

10.1016/j.jlp.2022.104779

1873-3352

Yiyue Chen, Laibin Zhang, Jinqiu Hu, Zeyu Liu, Kangkai Xu,

Offshore Engineering and Technologies

Long-distance oil and gas pipelines undergo great changes in natural environments and geographical conditions along length. Inadequate communications provided limited and vague information about accidents. Scattered emergency resources need adequate response preparations to reduce recovery time after an accident. Examining historic similar cases in the database in order to propose effectual accident responses is worth studying. In emergency response case-based reasoning (CBR), information on new accidents, such as type, degree, hazard object, etc. need to be as detailed as possible. The anticipated quality and efficiency of traditionally case recommendations might have been directly negatively influenced, by the delayed and ambiguous information available from new pipeline accidents. Thus, to fully extract the required information, and to minimize limitations on reasoning caused by a lack of accident information, this study proposes an accident Case Representation Model. The model contains: (i) an accident features dataset featuring a bipartite graph from hundreds of pipeline cases, and (ii) cases vectorization obtained by applying graph representation learning. As a result, (iii) recommendations of any new accident can be given based on similar case responses. In order to examine the accuracy of the recommended results, a relative similarity concept is proposed, showing that the proposed model's recommendation accuracy is basically maintained at 96.7%. Compared to other case vectorization methods, the accuracy increased by 4%–23%. Taking a block valve failure accident as a case study, this model expands on two related scenarios: overpressure in upstream pipeline and interruption of communication system. The supplementary responses increase the preparation comprehensiveness.