Search

The aim of the current work is to reproduce spills of fuel from storage tanks in order to better understand and replicate the effects of this phenomenon which can cause several damages. The vapour production that occurs when there is a failure in a storage tank could cause an explosion if there is any source of heat such as an electrical power failure. The vast increase in computational resources that we experienced in the last years has given the Computational Fluid Dynamics (CFD) community growing resources to simulate difficult flow problems that were impossible to solve 20 years ago. In order to solve the multiphase flow which represents the spilling of fuel from a storage tank a Lagrangian-Eulerian approach is adopted, as with most of the work done in the literature. The importance of splashing of droplets on at surfaces has also led to the development of an adapted splashing model based on existing correlations. This was done by the use of a volume-of-fluid (VOF) methodology to characterise splashing. Comparison of experimental results with the ones obtained by numerical simulation show good agreement (less than 10% error) and confirm that CFD could be an advantageous tool in the prediction of this type of flow. Moreover, the computational simulations give an overview of what is happening and more specifically physical quantities in each point of the computational domain, while the experimental facilities are restricted to some points and the presence of some tools invade the flow. The main findings of this work are related to the splashing of liquid droplets into solid surfaces and the differences between cascades of hydrocarbons that have a boiling point above normal ambient temperatures and liquids such as Liquefied Natural Gas (LNG) that are boiling when they come in contact with the atmosphere. The splashing of liquid droplets in the current application was found to be significantly different respect to most of the work discussed in the literature. Because of the large size of the droplets simulated, the parameters of interest such as splashing threshold, angle and mass splash ratio were found to be out of the previous ranges predicted by other authors. These parameters were then put back into the splashing model which showed improved results, giving a higher accuracy regarding vapour produced. Liquid cascades of Liquefied Natural Gas were found to be significantly different than gasoline ones, with most of the liquid being vapourised before hitting the ground. This is due mostly to the boiling of the liquid and also on the different layout of the storage tanks, these being much larger in an LNG plant than in a typical fuel plant. The models formulated in the current application can be used for the prediction of the flammable cloud in a spill scenario in a plant, where the vapour can reach high concentration and the risk of an explosion is not remote. The physical models mentioned above were implemented within the framework of the opensource tool OpenFOAM by modifying and/or adding new models in the code.