Leakage analysis and prediction model of underground high-pressure natural gas pipeline considering box culvert protection

Zhiheng Xia,Zhao-Dong Xu,Hongfang Lu,Haoyan Peng,Zicong Xie,Yankun Jia,Hui Sun
DOI: https://doi.org/10.1016/j.psep.2023.10.052
2023-11-02
Process Safety and Environmental Protection: Transactions of the Institution of Chemical Engineers, Part B
Abstract:Buried gas pipelines frequently experience failures as a consequence of adverse factors. The studies of gas leakage diffusion behavior and prediction of hazardous areas following pipeline failure hold profound significance in mitigating the risk of accidents. This paper exemplifies a specific segment of long-distance transportation gas pipeline in the "West-to-East Gas Transmission" project, employing the Soave-Redlich-Kwong (SRK) equation of state (EOS) to develop a numerical model for high-pressure gas leakage and diffusion, while considering box culvert protection. The characteristics of gas leakage diffusion and the effects of pressure, leakage diameter, and soil type on gas diffusion are presented. Furthermore, the prediction models for leakage rate, fastest hazardous time (FHT), fastest hazardous distance (FHD), and lateral hazardous distance (LHD) are introduced. The results show that the pressure and velocity at the beginning of the leakage manifest an annular diffusion centered on the leakage hole, stabilizing within 300 s. The gas primarily diffuses along the horizontal direction within the box culvert, and the peaks of gas concentration at the surface are situated at a distance of 2.35 m on either side of the pipeline. Variations in pressure and leakage diameter influence the gas concentration but do not alter the locations of the concentration peaks observed at the surface. The greater the viscous and inertial resistance of the soil, the shorter the overall diffusion distance becomes. However, backfilling the loam within the box culvert significantly elevates the concentration of gas that leaks to the surface. The leakage rate, FHT, FHD, and LHD can be well calculated by the prediction models proposed in this study. Graphical Download : Download high-res image (154KB) Download : Download full-size image
What problem does this paper attempt to address?