Development of a model for computational fluid dynamics simulation of liquefied natural gas vapour dispersion

Udechukwu, Izunna David (2015) Development of a model for computational fluid dynamics simulation of liquefied natural gas vapour dispersion. (PhD thesis), Kingston University, .

Abstract

Liquefied Natural Gas (LNG) is currently playing an important role in the world energy markets. This is evidenced by growing demand and increased construction of LNG facilities across Europe and the United States. In the event of spill from any of the facilities handling LNG such as during liquefaction, transportation or regasification, flammable vapour is formed which disperses through the atmosphere constituting fire and explosion hazards. To ensure public safety in the midst of growing LNG demand and facilities construction, industries are usually mandated to demonstrate that public safety will not be undermined by potential spill from their facilities. One method that is currently being used to demonstrate compliance is through LNG vapour dispersion modelling using Computational Fluid Dynamics (CFD). CFD modelling of dispersion phenomena is a challenging task that requires rigorous methodology to account for the underpinning physical processes. The modelling process comprises of two steps: source term quantification and vapour dispersion modelling. Source term quantification involves the physical description of spill rate, pool spreading and evaporation. Vapour dispersion utilizes the result of source term quantification in order to predict the turbulent entrainment and dilution process with the ambient wind. Existing models employ simplifying assumptions that circumvents explicit source term modelling. The spilled liquid is assumed to fill the entire substrate immediately at which time the spill rate becomes equal to evaporation rate. Following this assumption, a fixed inlet patch area and evaporation rate is applied at the gas inlet boundary. This approach fails to incorporate the transient pool development and subsequent evaporation into the dispersion modelling process. The primary aim of this dissertation is to develop an efficient integrated pool spreading, evaporation and dispersion (I-PSED) model code for LNG vapour dispersion simulation. This represents a significant shift from the traditional method since the new methodology combines the spilling process, spreading on substrate and transient evaporation into a unified model. For the spilling process, the well- known orifice model has been adopted to predict the spill rate taking into account the decreasing head. A mass balance approach is adopted in conjunction with a well¬established similarity model for spreading calculation. Heat transfer to the spreading pool is incorporated based on film boiling correlation. The spreading model was then coupled to an atmospheric dispersion model within OpenFOAM framework through the implementation of a new boundary condition in which the gas inlet patch area changes based on the instantaneous pool radius. The developed integrated code (I-PSED) is validated against data from the Coyote Series LNG Spill experiments as well as against Shell's Maplin Sand LNG spill experiments. Predictions of concentration obtained using the proposed model and those obtained using conventional approach are compared against experimental data at specific sensor locations. Also, arc-wise comparisons are carried out. Predicted results show good agreement with experimental data and clearly put the newly developed model ahead of the conventional approach for CFD simulation of LNG vapour dispersion. With the newly developed approach, the cloud arrival time and average concentrations at most sensor locations were better predicted. The effect of the turbulent production due to density stratification (buoyancy) created by the release of cryogen is investigated. Experience gathered shows that incorporation of a production term due to buoyancy in the turbulence model improves predictions under unstable atmospheric condition, otherwise the concentration field would be grossly over-predicted.

Actions (Repository Editors)

Item Control Page Item Control Page