After the high-profile accidents at Buncefield and Texas City, risks from gas explosions have been high on the agenda. The need for transparency is greater than ever; particular focus has been on occupied buildings and the protection they offer their occupants. The explosion extension planned for Phast 6.6 provides new functionality supporting increased accuracy and transparency.
When looking at risks to people from gas explosions, it is necessary to predict the behaviour of the gas cloud, its interaction with process plant, the probability it will ignite, the overpressures generated if it does ignite, and the effects of these overpressures on the population.
In Phast Risk, the calculation sequence in time is discharge, dispersion, ignition, explosion, damage and lethality. The work described here focuses on the last three stages.
Two of the most widely used models for calculating overpressure from gas explosions are the Multi-Energy Method and the Baker-Strehlow-Tang Model (BST).
The former is preferred in Europe, where it was developed as a more accurate alternative to the TNT-equivalence model previously used for simplicity in QRA. The latter, which relates overpressure to distances from the explosion centre for different equipment density, is the preferred model for the assessment of risks from explosions in the Americas. Both models are being included in an extension to Phast Risk. Since these models were first proposed, a number of industry-funded initiatives have been taken to improve their accuracy and usability, the findings from which also being taken account of in the software.
A technical paper on this topic will be published at the Hazards XXI conference in Manchester, 9–12 November. Due for completion at the end of 2009, this work will be available in Phast Risk 6.6. Overpressures are calculated using either model, both of which take account of surrounding plant in line with a number of industry guidelines. Risk calculations also take into account buildings and the protection they offer their occupants.
Because of the complexity of real process plants, it is not trivial to apply these models to represent typical accident scenarios. A particular issue is that different analysts may make different assumptions and obtain different results. This has led to research comparing the models with experimental results and more rigorous
models, such as CFD. From this research, guidance has been provided on how the models should be applied to ‘real plants’.
Between 1993 and 1995, a joint industry project sponsored by twelve organisations provided ‘Guidance for the Application of the Multi-Energy model’, commonly referred to as the GAME project. This research developed correlations for the maximum overpressures at the explosion centre for different types of equipment. The estimated overpressures can then be used for consequence and risk calculations.
However, determining the parameters to be used for selecting the best correlation from the GAME project is not a straightforward task. Another joint industry project, sponsored by eleven organisations, provided guidance on the practical application of the GAME recommendations to specific example scenarios. The Multi-Energy results were compared with detailed information provided by measurements and CFD predictions of the explosion behaviour. This work was published in 1998 under the project acronym GAMES. In parallel with this project the Dutch Yellow Book was updated, so some of the research and guidance was reflected in the 1997 version of this document.
An important aspect of applying the Multi-Energy and BST models is determining the location of explosion sources. This is difficult if a flammable cloud engulfs areas of plant separated by open spaces. If the open spaces are sufficiently large, the explosion flame front will slow down significantly while travelling across them,
and the explosion will develop multiple separate pressure waves. However, if the spaces are small, the explosion should be modelled as a single-pressure wave.
A third joint industry project, RIGOS, was funded to investigate this phenomenon. It included an experimental programme to provide insight into the influence of separation distance between areas of plant on the explosion behaviour. RIGOS made a number of recommendations on critical separation ratios that should be used based on explosion overpressure. These will be implemented in the software.
Once the overpressure from an explosion has been predicted, one needs to assess the effects on the population. Converting harmful effects into rates of fatality and injury is known as ‘vulnerability modelling’; there are a number of published vulnerability models for explosion effects. These are used to calculate the probability of death based on the results of the explosion models described above.
A number of methods from published literature are included in the software which allow individual buildings to be defined. This provides a much more accurate picture of the risks to persons inside buildings.