Validation of BLEVE events using the hybrid code exploCFD

Abstract

A hybrid (analytical-numerical) method for the modeling of BLEVE events seems the most realistic approach given the complex coupled physical models required for a purely numerical approach. The code exploCFD employs just such a method for the calculation of events involving multiple types of explosions including BLEVEs whereby the results from an analytical model are fed into a CFD code to model the wider domain. The CFD code is configured as a 2D solver hence the methodology carries some uncertainty. So how do the results from a hybrid method compare with those obtained from experiments? Here we look at the BLEVE experiments of Birk and Vandersteen. We show that despite simplifications inherent in the methodology, which include the combining separate discrete events such as initial tank rupture, catastrophic failure and mixing and subsequent ignition into one equivalent event, exploCFD produces results which are in good agreement with the experimental peak pressures and impulses. The peak pressure is captured at locations 10, 20, 30, and 40 m from tank side and the impulse values also show good agreement. Surprisingly, the time histories also show better than expected agreement.

CSC coupled Graham Kenney source term for hybrid models of BLEVE events: Formulation, implementation and validation

“… show comparisons between the first peak time histories of the experiments reported in Laboreur et al,19 and simulations using hybrid methodology carried out using exploCFD for sensor locations 10, 20, 30, and 40 m away from the side of the tank. It can be seen that the agreement is excellent, exceeding in accuracy previous attempts using hybrid methodology 10. When taking into account the signal drift as well as the high frequency noise seen in the experimental results, and which are shown in Figures4-7as vertical error bars, the exploCFD prediction for the maximum overpressure comes within a few hundred Pascals of the experimental readings.…”

“…Mesh independence has been extensively carried out for this methodology, including for BLEVEs in previous studies. 10 By contrast, software packages with sub grid models, there is usually a recommended cell size which cannot be changed. In those cases, mesh independence cannot be carried out.…”

“…It is unlikely that the fraction evaporated is a fixed percentage but is rather case dependent. In 2020, we proposed an approach based on the hybrid method 10 for the calculation of BLEVEs using on Prugh's method. The method was shown to capture the maximum overpressures and the impulse values with good agreement with Birk's experiments in some cases, but the agreement could clearly stand improvement.…”