Computational fluid dynamics for LNG vapor dispersion modeling: a key parameters study

The increased demand for liquefied natural gas (LNG) has led to the construction of several new LNG terminals in the United States (US) and around the world. To ensure the safety of the public, consequence modeling is used to estimate the exclusion distances. The purpose of having these exclusion distances is to protect the public from being reached by flammable vapors during a release. For LNG industry, the exclusion zones are determined by the half lower flammability limits (half LFL, 2.5% V/V). Since LNG vapors are heavier‐than‐air when released into atmosphere, it goes through stages, negative, neutral and positive buoyant effect. In this process, it may reach the half LFL. The primary objective of this dissertation is to advance the status of LNG vapor dispersion modeling, especially for complex scenarios (i.e. including obstacle effects). The most used software, box models, cannot assess these complex scenarios. Box models simulate the vapor in a free‐obstacle environment. Due to the advancement in computing, this conservative approach has become questionable. New codes as computational fluid dynamics (CFD) have been proven viable and more efficient than box models. The use of such advance tool in consequence modeling requires the refinement of some of the parameters. In these dissertation, these parameters were identified and refine through a series of field tests at the Brayton Firefighter Training Field (BFTF) as part of the Texas A&M University System (TAMUS). A total of five tests contributed to this dissertation, which three of them were designed and executed by the LNG team of the Mary Kay O'Connor Process Safety Center (MKOPSC) and the financial support from BP Global SPU Gas (BP). The data collected were used as calibration for a commercial CFD code called CFX from ANSYS. Once the CFD code was tuned, it was used in a sensitivity analysis to assess the effects of parameters in the LFL distance and the concentration levels. The dissertation discusses also the validity range for the key parameters