The Buncefield Explosion: Were the Resulting Overpressures Really Unforeseeable? Je ´ ro ˆ me Taveau Institut de Radioprotection et de Su ˆ rete ´ Nucle ´aire, IRSN/DSU/SERIC/BEXI, 31, avenue de la Division Leclerc, 92 260 FONTENAY AUX ROSES cedex, France; jerome.taveau@irsn.fr (for correspondence) Published online 14 July 2011 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/prs.10468 On Sunday 11 December 2005, a severe unconfined vapor cloud explosion followed by several tank fires occurred at the Buncefield oil storage depot in England, near London, caus- ing widespread damage to homes and businesses surround- ing the site, hopefully without any victim. The damage caused by the resulting blast wave(s) sur- prised all the process safety community and explosion experts, as common hazard assessments would have predicted over- pressures of only 5 kPa (0.73 psi), instead of 200 kPa (29 psi) as suggested by the data collected onsite. One of the particularities of the Buncefield oil storage de- pot is that it was surrounded by long and continuous rows of trees and bushes; this singularity is currently considered as a preponderant factor in the resulting high overpressures. Recently, explanations of the resulting overpressures were given by some authors using data from large-scale explosion tests and CFD simulations. The objective of this article is to verify whether application of simple methodologies used in common risk assessments could allow to estimate the severity of the Buncefield explosion. Ó 2011 American Institute of Chemical Engineers Process Saf Prog 31: 55–71, 2012 Keywords: Buncefield, explosion, overpressures, bang-box, Kinsella, Congestion Assessment Method, Multi-Energy Method, Baker-Strehlow-Tang method, MERGE, GAME, DDT INTRODUCTION On Sunday December 11, 2005, a severe unconfined vapor cloud explosion occurred at the Buncefield oil storage depot in England. The resulting damage was quite unusually severe (Figure 1) and surprised all the process safety community; as stated in the Third Progress Report of the Buncefield Investiga- tion Team: ‘‘The magnitude of the overpressures generated in the open areas of the Northgate and Fuji car parks is not con- sistent with current understanding of vapour cloud explosions. For example, a method in current usage would predict over- pressures in this sort of environment of 20–50 mbar. The inves- tigation has, so far, been unable to establish why the ignition of the vapour cloud and the explosion propagation in the rela- tively uncongested environment of the adjacent car parks caused significant overpressures that produced the severe dam- age to property.’’ Trees and bushes surrounding the oil depot were pointed out to be responsible for the resulting overpressures, as they provided a sufficiently congested area for strong flame accel- eration. Several authors have presented their own interpretation of the resulting overpressures by using large-scale explosion tests data [2] or sophisticated computational fluid dynamics (CFD) simulations [3,4]. Unfortunately, large-scale explosion test data are often confidential, and CFD simulations of such large-scale accidents are too time consuming for common process safety engineers. In this context, a remaining question is whether sim- ple methodologies enable to correctly estimate the resulting overpressures. This article tries to address this issue by applying four well-known methods in order to estimate a posteriori the maximal overpressures consecutive to the Buncefield explo- sion: • the initial blast strength index proposed by Kinsella [5]; • the Congestion Assessment Method (CAM) developed by Cates [6]; • the Baker–Strehlow–Tang method set up by BakerRisk [7]; • the GAME correlation established by The Netherlands Or- ganization for applied scientific research (TNO) [8]. Results and inherent limitations of such methods will then be reviewed. Also, the explosion scenario proposed by the Buncefield Explosion Investigation Team will be discussed. THE BUNCEFIELD EXPLOSION Sequence of Events On Saturday December 10, 2005, at around 07:00 pm, tank 912 of the Buncefield oil depot (Figure 2) started receiving unleaded motor fuel from the Coryton refinery [10]. From approximately 03:00 am on Sunday December 11, the level gauge of tank 912 recorded an unchanged reading (two-thirds filling level). However, filling of tank 912 contin- ued. Calculations show that tank 912 started to overflow at around 05:20 am, causing the fuel to fall down on the side of the tank and through the air. This led to the rapid formation of a rich fuel–air mixture (Figure 3). The vapor cloud started to spread on the Buncefield oil depot, and then offsite, to cover lastly a total area of 120,000 m 2 , with an average height of about 3 m. On Figure 4, the maximal extension of the flammable cloud corresponds to the area affected by scorching. Finally, a big explosion occurred at 06:01 am (40 min after the tank overfilling began) followed by further explosions and tank fires. Ó 2011 American Institute of Chemical Engineers Process Safety Progress (Vol.31, No.1) March 2012 55