Numerical modelling of inhomogeneous Liquefied Natural Gas (LNG) vapour cloud explosions

Khodadadi Azadboni, Reza (2019) Numerical modelling of inhomogeneous Liquefied Natural Gas (LNG) vapour cloud explosions. (PhD thesis), Kingston University, .

Abstract

The main hazard of Liquified Natural Gas (LNG) is the flammable vapour cloud, which can extend to kilometres as a greenhouse gas or be ignited resulting in fire and explosions. This work aims to carry out a safety study on the vapour cloud explosion of LNG. Since most of the experimental research works are available for Hydrogen/Air mixture, in this present work, the first goal is to extend the existing physical understanding of deflagration-to-detonation transition (DDT), in hydrogen/air mixtures with transverse concentration gradients in closed channels. Explosions in homogenous (uniform) reactive mixtures have been widely investigated, both experimentally and numerically. However, in real accident scenarios, mixtures are usually inhomogeneous due to the localised nature of most fuel releases, buoyancy effects and the finite time between release and ignition. It is imperative to determine whether mixture inhomogeneity can increase the explosion hazard beyond what is known for homogeneous mixtures. Hence, extensive knowledge on these processes has been built up over decades for homogeneous mixtures. The approach is to identify similarities and differences caused by concentration gradients compared to homogenous mixtures with equal average hydrogen concentration. The dynamics of deflagration to detonation transition (DDT), and explosion modelling, have been studied using the newly assembled density-based solvers (VCEFoam) within the frame of OpenFOAM CFD toolbox. In order to evaluate the convective fluxes contribution, Harten–Lax–van Leer–Contact (HLLC) scheme is used for accurate shock capturing. The numerical code is initially verified by four sets of verification test cases. In addition to shock capturing verification, the capability of the current numerical code in capturing the detonation cellular structure has been examined. The CFD results have been compared against both quantitatively and qualitatively with the other previous works as well as an experimental observation. Then, numerical studies have been conducted to investigate flame acceleration and transition to detonation in both homogeneous and inhomogeneous hydrogen-air mixtures in obstructed and unobstructed channel configurations (in medium scale). The developed VCEFoam solver has been used within OpenFOAM, for these simulations. For the considered experiment (Boeck et al., 2016), different sets of configurations and fuel concentration have been studied. Three different geometry configuration such as BR00 (0% Blockage ratio, smooth channel), BR30 (30 % blockage ratio), and BR60 (60% blockage ratio), have been considered in this study. Also, in order to study the effect of a concentration gradient, different mixture concentrations have been investigated in both homogenous and inhomogeneous mixtures. A total of 17 conditions were simulated for different hydrogen concentrations in both homogeneous and inhomogeneous mixtures with and without obstructions. A high resolution grid is provided by using adaptive mesh refinement (AMR) method, which leads to 30 grid points per half reaction length (HRL). The numerical predictions were compared against previous experiments. Overall, the predicted flame tip velocities, overpressures, and locations of detonation onset are in good reasonably agreement with the measurements. It is found that, the transverse concentration gradients can either strengthen or weaken flame acceleration, depending on average hydrogen concentration and channel obstruction. The role of hydrodynamic instabilities and the effect of baroclinic torque and Richtmyer Meshkov (RM) instability have also been studied. The results support that RM instability is the primary source of turbulence generation in the present case. Then vapour cloud explosion study has been carried out for industrial scale scenarios (very large scale). A robust CFD methodology has been presented for modelling very large scale, vapour cloud explosions scenarios. A specific model has been considered for modelling the impact of flame-instabilities, particularly the thermal diffusive instabilities, and Darrieus Landau (DL) instabilities in large-scale models. The numerical model has initially been validated with the largest ever conducted indoor DDT and explosion experiments in the RUT facilities. Good qualitative agreement between the numerical prediction results and experimental measurements of RUT facilities has achieved. After demonstrating the code verification, LNG vapour cloud explosion scenarios, generated from the release of an evaporated liquefied natural gas have been studied. Two different possible incidents in LNG VCE have been studied; explosion modelling in onshore LNG plant and offshore LNG shipping. For the onshore LNG explosion study; an LNG plant has been considered to have fuel leakage from one of its storage tanks. In both onshore and offshore scenarios, the maximum recorded overpressure was below 1.2 bar, which is far below the CJ detonation limit (CJ detonation pressure, for stoichiometric methane/air mixtures, is 16.6 bar). Therefore, in this scenario, LNG flame acceleration was not enough to cause a detonation, and only a flame deflagration has been noticed. The results of the current study can be used in the context of safety to assess the potential risks of explosions in the energy industry.

Actions (Repository Editors)

Item Control Page Item Control Page