Predictive model of bund overtopping fraction for catastrophic failure of storage tank. (September 2020)

• Record Type: Journal Article
• Authors: Luan, Xiaoyang; Huo, Jiali; Liu, Yingchun; Jiang, Juncheng; Zhang, Bin
• Is Part Of: Safety science. Volume 129(2020)
• Publication Date: 2020-09
• Subjects: Storage tank -- Catastrophic failure -- Secondary containment -- Bund overtopping; Industrial accidents -- Periodicals; Accident Prevention -- Periodicals; Safety -- Periodicals; Travail -- Accidents -- Périodiques; 363.11
• DOI: 10.1016/j.ssci.2020.104801
• Languages: English
• ISSNs: 0925-7535

Abstract

Highlights: Three novel models were developed to predict the overtopping fraction of catastrophic tank failures with circular bunds, square bunds and bunds of arbitrary shapes. The inclination angle ( θ ) was first introduced into the predictive model of overtopping fraction. The novel models were validated against existing experimental data to verify their performance. Abstract: The storage tanks are usually equipped with a secondary containment, such as a dike or a bund, to control the spilled liquid after a tank failure. The bund system is extremely important for storage tanks of hazardous materials, such as Liquefied Natural Gas (LNG), because the spilled LNG tends to vaporize rapidly with a low boiling temperature, and the bund contributes to reducing the vaporization rate by controlling the contact area of conductive heat transfer. Some regulations mandate the 110% rule for the bund design; however, a bund that complies with this rule may not hold all the fluid after a catastrophic tank failure, and some fluid would overtop from the bund wall to escalate the incident. Previous works have developed several predictive models of overtopping fraction for circular bunds, but these works only used a small set of experimental data, did not include the parameter of inclination angle, and failed to verify their performance for bunds of other shapes. This work collected all reported data of experimental studies on bund overtopping, and used them to develop novel predictive models of overtopping fraction for circular bunds, square bunds and bunds of arbitrary shapes. These predictive models were evaluated using a number of statistical performance measures. The findings were used to provide recommendations on the bund design and the future work.