Vapour cloud explosion incidents

Review of Vapour Cloud Explosion Incidents, G. Atkinson, J. Hall, A. McGillivray, Research Report 1113, HSE Buxton, 2017.

Vapour Cloud Explosions an important new safety case reference

Anyone involved in compiling, reviewing or updating a safety case for an installation where large petroleum vapour or gas releases are possible should study this valuable, free HSE Research Report, published July 2017. The report describes work done by HSE with US safety regulators to consolidate previous research and to incorporate recently published analysis into a single, systematic review of historical incidents.

Even for installations operating in non-Safety Case regulatory regimes, both on- and offshore, there is much for both designers and operators to learn and consider. This major report (320pp) provides full details, including graphic photos, of five major incidents (in India, UK, Venezuela and USA), plus summaries for 19 others because only less detailed records are available see Section 3.2.

The joint review was initiated to better understand vapour cloud explosion risks at LNG plants. But many of the findings are also relevant to other types of facility including: gasoline storage depots, tanker terminals, refineries and chemical processing sites where hydrocarbons in the range C2-C6 are present. There are also implications for the assessment of risks from pipelines.

If you know little about how explosion events escalate from small beginnings to a major disaster, Section 4 of the report provides clear explanations, so is a valuable training resource. If you already know a good deal, you should review your spill/leak detection and mitigation barriers in the light of its conclusions, and will probably wish to update relevant facility risk profiles.

The key findings include:

  • A high proportion of the incidents (~ 60%) occurred in nil/low wind conditions, even though such atmospheric conditions are present typically for only 5% of the time.
  • For these events, the initial leak/spill was typically not detected, so the part of the release within the flammable zone grew until it reached a distant ignition source.
  • In contrast, where a release is more diluted with air, due to either its own momentum or wind effects or, most likely, a combination of both, if it ignites the resulting overpressures are much less damaging.
  • Area gas detection is often not specified for open areas because, under typical weather conditions, a very large number of detectors would be needed for reliable detection of likely releases. However, reliable detection in nil/low wind conditions may be more feasible, thus significantly reducing risks from the most damaging events.
  • In more recent explosions, forensic examination of damaged items has revealed key details about actual overpressures, event timescales, etc. Such data were not recorded in most historic explosion investigations but can help to validate current design models, risk assessments, etc. So, if you investigate even a minor explosion event, make sure suitable expertise is deployed to maximise both technical and practical lessons learned.
  • The most damaging vapour cloud explosions (detonations) occur at high flame speeds. Explosions with low flame speeds (deflagrations) generate lower overpressures, and thus less damage. The detailed evidence in this review is not fully consistent with current models for the transition between the two categories, so continuing research is needed in that area.

These findings may seem to be largely design-related, and thus fairly irrelevant for operations at an existing facility. But its vital that operations personnel have a good understanding of how bad a worst-case release and ignition could be, because they use and maintain the various barriers that prevent leaks and spills, detect them as soon as possible and then respond to the emergency. In particular, if any of these barriers become unavailable or ineffective, they have to decide whether normal operations can continue. The photos of real events in Section 5 and Appendices 1 and 3 can be used to prepare them for such decision making.

Vapour cloud explosions are technically complex events, so some of the details may be too abstruse for some readers. But, overall, this is a really important publication with a wide range of uses for designers and operators in organisations handling large volumes of petroleum gases and liquids, and for the OSH professionals who advise them. Above all, its free to download, though quite large (~12MB).

Ian Waldram, CFIOSH