(65as) Apparent Discrepancies in Methods for Predicting the Explosion Energies of Bleves

The authors have used calculation methods for BLEVE blasts as described in the CCPS book Guidelines for Vapor Cloud Explosion, Pressure Vessel Burst, BLEVE and Flash Fire Hazards, and most recently similar methods in the PHAST process industry hazard analysis/consequence analysis software tool, developed by DNV-GL. Predictions made using PHAST appeared to be overly conservative, which instigated a comprehensive comparison of PHAST to other methods in the literature. There is not universal agreement concerning the dominant mechanisms driving the air blast pressure wave from a BLEVE, and there are major differences in methods used to predict the explosion energy driving the blast wave.  The CCPS guidelines book and PHAST consider the explosion energy driving the blast wave to be from both the vapor and liquid phases present at the time of loss of containment. For each phase the contribution to explosion energy is the difference in internal energy between the thermodynamic states of the liquid and vapor at the time of loss of containment and the final states of the materials when isentropically expanded to atmospheric pressure.  Recently, Birk compared various prediction methods to experimental data for propane BLEVEs and concluded that the flash vaporization of the liquid may be too slow to contribute significantly to the blast, therefore the blast energy is primarily from the expansion of the vapor in the vessel at the time of loss of containment.  Other authors also base the explosion energy on the gas expansion and others on the liquid superheat. Ogle and coworkers have discussed the available energy of a BLEVE (exergy) and compared this to the BLEVE explosion energy predicted using several methods in the literature.   These various approaches can lead to drastically different predictions of explosion energy and the resultant blast pressures and impulses.  BLEVE theories and models are discussed in this paper and predictions are compared to limited experimental data. This paper is meant to provide comparisons between the many various approaches, including many more that are not listed above, and summarize their differences.