Comprehensive evaluation of landslide risks of oil and gas pipelines based on cloud theory

Landslides are serious geological hazards along long-distance oil and gas pipelines. Especially common are discontinuous-developing single landslides. A single landslide hazard can cause anything from pipeline rupture and fracture to complete failure and shutdown, thus triggering serious secondary disasters. Risk assessments of oil-and-gas-pipeline landslides are an effective method for ascertaining the degree of landslide risk and can provide an important scientific basis for planning and decision-making regarding landslide prevention and control along long-distance oil and gas pipelines. In addition, risk assessments represent an important step in the pipeline-integrity management process. The evaluation system consists of both quantitative and qualitative indexes, which are characterized by randomness and fuzziness. To address the subjectivity and incompleteness of qualitative and semi-quantitative evaluation methods in the processing of randomness and fuzziness, the cloud theory was introduced, which can simultaneously reflect randomness and fuzziness. The golden section method was used to establish a five-level standard cloud metric for pipeline landslide risk and index weighting. In the cloud transformation process, this paper proposes uncertainty reasoning for the quantitative index and a floating cloud preference algorithm for expert group language as a qualitative index, which comprises the assessment model for landslide risk of oil and gas pipelines. The comprehensive evaluation results indicate that the floating cloud preference algorithm for the qualitative index is more suitable for the language of expert group decision-making than the synthetic cloud algorithm commonly used. In addition, the results of the four pipeline-landslide-risk evaluations are basically consistent with the results of the semi-quantitative method, which is consistent with the actual situation. This method softens the hard divisions between the inner boundaries of the index and simplifies the preprocessing of index data. It fuses the qualitative and quantitative evaluation aspects using composite decision-making and improves the accuracy, rationality, and visualization of the results.