Worst case explosion analysis
The objective of a worst case analysis is to assess the maximum realistic overpressure levels that can be experienced with a specific layout. GexCon recommends that this be performed as follows:
- Perform a limited number of dispersion simulations, where the objective is to try to produce clouds with as large as possible combustible parts (looking for worst case results)
- Perform a limited set of explosion simulations based on the results of the dispersion simulations, varying cloud size and ignition location
It is possible to simplify this task by just assuming cloud sizes for the explosion analyses, in which case the cost will be lower but the results more uncertain.
In an explosion simulation based on dispersion analysis, the relevant amount of gas to use is what is close to stoichiometric. The figure below illustrates a 2D snapshot of a dispersion simulation as well as the time development of the total amount of gas, the amount between the combustible limits and the amount close to stoichiometric, all in a specified area.
Note that the major parameters influencing the overpressure in a highly congested region are confinement and amount of gas-air mixture close to stoichiometric concentration. It is therefore of paramount importance that representative cloud sizes are used in an explosion analysis. Hence it is recommended that the cloud size evaluation is based on CFD simulations (as proposed above), also accounting for the geometry congestion.
Experience indicates that gas clouds (i.e. the combustible part of a cloud) may typically cover a much smaller area than an entire deck or module, see an example in the figure below.
It is possible to do a simple study with few simulations. The problem with this is that the question of representativity becomes an issue: the term "a representative cloud" will then have to mean "a cloud which may occur". We do not have enough information to map the possible size distributions. It will not be possible to determine the degree of conservatism of using a very small set of clouds, and one will hence know nothing about the resulting overpressures: are they on the high side, low side, or somewhere in between? There are really only two acceptable approaches: do a simple but assuredly conservative analysis, or do a more extensive, probabilistic analysis. In many cases, however, the simple conservative analysis will result in overpressures which are too high to be manageable. Hence the probabilistic analysis may be the only viable solution. This point of view is also consistent with the requirements of the ISO 13702 standard for fire and explosion protection offshore.
Explosion simulations performed in a worst case or a probabilistic analysis may be used as a basis for a more detailed assessment of the response of specific items e.g. pipes, vessels, valves, cable trays or critical structural members. The simulations will be redone with detailed monitoring of overpressure, impulse, drag and drag impulse distribution close to the selected items. The results will be used to produce graphs of equivalent static loads for different support conditions. Guidance will be provided on the use of the information for structural analysis of supports.
This methodology has been developed in an HSE-sponsored study, looking at the effect of explosion loading on equipment for three platforms.
Contact
Lars Rogstadkjernet
Phone no: +47 55 57 43 22
Email: larsr(at)gexcon.com
