Investigation on the concentration prediction model and personnel hazard range of LNG leakage from tankers in the tunnel

Abstract

Liquid natural gas (LNG) leaks from tanker in the tunnel could result in more severe fire and explosion catastrophes than those that occur in unconfined space. Investigating the gas concentration distribution law and the accompanying personnel hazard range is crucial to provide guidelines for personnel escape and fire rescue in such accidents. In this paper, to provide a rapid technique for concentration estimation, the virtual source approach and the fluid settlement theory are applied to put forward a new model, which is based on the Gaussian plume model. Three batches of LNG diffusion tests were carried out in a 5 m long wind tunnel with a cross-section of 1 m × 1 m. The comparison results show that the prediction accuracy of the proposed model in the far region of the leakage source is better than that in the near region, with the error within ±16%, which demonstrated an accurate prediction performance of the proposed model. The numerical model of LNG leakage from tanker in the tunnel is established, and the influence of leakage direction, leakage aperture, and tunnel cross-section on concentration distribution law is discussed. The findings indicate that when the leakage direction changes from horizontal to vertical, the gas is more likely to accumulate in the tunnel due to the greater attenuation of the initial momentum of the gas, which causes the combustible range to increase by about 2 times. When the leakage aperture is 50 mm, the mass of combustible gas reaches the maximum value of 320.91 kg in 85.9 s, which indicates that the explosion intensity is not proportional to the leakage aperture. Excessive leakage aperture causes the concentration to surpass the Upper Flammable Limit, and the similar trends occur when the tunnel cross-section reduces. Based on the numerical results, the calculation model of the personnel hazard range suitable for various tunnel cross-sections and leakage apertures is derived by data fitting.