A Proposal for Bayesian Handling of Information on Dimensions of Fragments Generated by Fire-Induced Explosions of Cylindrical Vessels

27 Pages Posted: 16 Jul 2024

Abstract

The problem of assessing dimensions of fragments generated by fire-induced explosions of cylindrical pressure vessels is considered. The attention is focussed on boiling liquid expanding vapour explosions (BLEVEs). Assessing fragment dimensions comprises acquisition of raw data, bringing/processing this data to the form suitable for a Bayesian analysis and application of processed information to predicting fragmentation of vessels in future BLEVE accidents. The proposed procedure of assessing fragment dimensions is based on our previous work, in which a Bayesian approach to predicting fragmentation/failure patterns of cylindrical vessels has been developed (Vaidogas, 2021a). Algorithms suggested in the present study for collecting and processing data on fragment dimensions are conditioned by an existence of three fragmentation patterns that prevail in the statistics of vessel ruptures and produce rocket-like fragments known as oblong end-caps. These fragments are viewed as be the most dangerous products of vessel ruptures. Assessing fragment dimensions is reduced to dealing with the length of the oblong end-caps. Other primary fragments, such as flattened parts of vessel shell or relatively flat portions of vessel heads, are rated less hazardous and therefore are not included in the analysis. Processing data on the length of oblong end-caps is based on transforming raw (in-the-field) measurements into relative lengths belonging to the unit interval. Uncertainty in the relative length of the oblong end-caps is modelled by means of beta and Dirichlet distributions. It is shown how to update these distributions with new evidence consisting of the relative fragment lengths. Finally, the study suggests two algorithms developed for recording and processing data on the length of oblong end-caps in the course of post-mortem investigations of BLEVE accidents. These algorithms are supplemented with one further algorithm intended for simulating (repeated Monte Carlo sampling) the dimensions of the oblong end-caps. It is suggested to use output of Monte Carlo sampling for generating masses and ejection velocities of fragments produced by BLEVEs. Results this study have been developed with the aim to improve assessing and reducing the risk posed by potential impacts of fragments generated by fire-induced vessel ruptures.

Keywords: BLEVE, fragmentation pattern, fragment, oblong end-cap, Bayesian updating

undefined

Suggested Citation

Vaidogas, Egidijus Rytas, A Proposal for Bayesian Handling of Information on Dimensions of Fragments Generated by Fire-Induced Explosions of Cylindrical Vessels. Available at SSRN: https://ssrn.com/abstract=4896046 or http://dx.doi.org/10.2139/ssrn.4896046

Egidijus Rytas Vaidogas (Contact Author)

Vilnius Gediminas Technical University ( email )

Lithuania

0 References

    0 Citations

      Do you have a job opening that you would like to promote on SSRN?

      Paper statistics

      Downloads
      18
      Abstract Views
      60
      PlumX Metrics