The growth of liquefied natural gas (LNG)’s importance for curbing greenhouse gas emissions has increased the interest in understanding LNG’s risks, particularly regarding small-diameter leaks (<25 mm). Recently, INERIS performed a series of pressurized LNG release experiments for orifice sizes of up to 9 mm. Based on INERIS findings and the Isenthalpic Homogeneous Equilibrium Model (HEM), this paper created a leak model for flashing LNG leak. The leak model consists of nine equations and quantifies leak parameters for risk assessment. One potential use of this leak model is providing an equivalent leak boundary condition for computational fluid dynamic (CFD) simulation to predict gas dispersion. Using the leak model as input, FLACS gas dispersion simulation was carried out for one INERIS experiment leak case. Compared to TR56, the Singapore safety guideline for LNG bunkering, the dispersion result does not contradict the expected plume reach. Further validation with risk analysis tools and actual experiments is needed to confirm that the leak model is fit for use.

No abstract

Performing liquefied natural gas (LNG) bunkering involves the risk of accidental leakage. When released from containment, LNG rapidly vaporizes into flammable natural gas and could lead to flash fire and explosion. Hence, LNG bunkering needs to take place in an area without an ignition source called a safety zone. This study compares the safety zone estimated by the Bunkering Area Safety Information for LNG (BASiL) model with that of the computational fluid dynamic (CFD) software FLACS, for Ro-Pax ferry bunkering. Horizontal leaks covering different wind speeds in eight wind directions were compared between the two models. Additionally, a grid refinement study was performed systematically to quantify the discretization error uncertainty in the CFD. Of 24 leak cases, FLACS and the BASiL model results agreed on 18 cases. In three cases validation was inconclusive due to the CFD error uncertainty. The BASiL model underestimated the safety zone distance in three cases compared with FLACS. Future work would be to perform a higher grid refinement study to confirm inconclusive comparison and examine ways to reduce gas dispersion spread for the worst result.