Parametric study of cryogenic jet leak phenomenon for risk assessment

• Main Author: Lim, Boon How
• Other Authors: Ng Yin Kwee
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Aeronautical engineering
• Online Access: https://hdl.handle.net/10356/170975
• Institution: Nanyang Technological University

The World Energy Outlook 2022 report forecasts that alternate fuels such as natural gas and hydrogen will meet approximately 20% of global energy demand from 2022 to 2030. These fuels are stored at cryogenic temperatures and vaporise upon leaving containment. The vaporised liquid can lead to fire and explosion, resulting in loss of life and asset damage. In June 2022, a liquified natural gas leak from a safety valve caused an explosion and rendered America Freeport’s export facility inoperable. As such, the adoption of alternate fuels brings new risks to society. Fuel risk leaks can be studied with computational fluid dynamics simulation by predicting gas dispersion. However, a pressurised jet leak is computationally expensive and time consuming to simulate. Furthermore, such leaks can occur in two phases, which increases simulation complexity. Thus, this thesis report addresses this challenge by accomplishing two objectives. The primary objective is to propose a pseudo-model to bypass jet simulation by providing an equivalent leak boundary condition for gas dispersion simulation. The secondary objective is to study three parameters – impingement distance, leaking pressure and ambient weather impact – on the two-phase fraction. Knowing the two-phase fraction is necessary for gas dispersion simulation. To validate the pseudo-model’s accuracy and perform a parametric study, experimental data is needed. Literature on two-phase leak data is sparse; therefore, a cryogenic leak experiment was performed. A total of nine liquid argon were leaked from a 10 mm-diameter orifice with varying pressure, impingement distances and weather conditions. Two additional liquid nitrogen leaks were conducted as well. Argon concentration levels relative to air were measured for pseudo-model validation. The leakage liquid fraction, known as rainout, was measured for the parametric study. In addition, wind velocity, ambient atmospheric pressure, ambient temperature, solar radiation and humidity were measured. The validation was conducted by simulating the experiment leak scenario with FLame ACceleration Simulator and a pseudo-model. Fractional bias, normalised mean square error and factor of two observations were used to evaluate the gas dispersion accuracy. The pseudo-model scored a fractional bias score of −0.257, which indicates an average overprediction of 29.4% for 13 sensor points. The normalised mean square error score is 0.354 demonstrating that the overprediction trend is consistent for all points. For the furthest-placed sensor, the prediction matches the experimental data. In short, the pseudo-model can predict far-field gas dispersion well. The parametric study revealed that impingement distance most significantly impacts the rainout phenomenon. The highest liquid fraction measured from an impinged leak was 0.63. No rainout was collected for non-impinged leaks. The observation was that a shorter impingement distance led to more rainout collection. Finally, the study suggested that relatively high humidity plays a role in the rainout phenomenon. This is because the vapour cloud formation, which is dependent on relatively high humidity, hinders heat ingress from the surrounding area and increases rainout quantity. A computational fluid dynamics simulation of leaks with and without impingement was performed. The results show that impingement can reduce the gas dispersion distance by as much as 71% (43.9 to 75.07 m) for a liquified natural leak from a ferry. This is because impingement reduces the volume of gas dispersion by generating rainout. Furthermore, the impingement can channel the dispersion away from the critical zone and enhance safety for society.