TABLE OF CONTENTS Ignition probabilities are one of the most critical elements of risk quantification in that the risk results are normally directly dependent on the probability of ignition. There are limited accident statistics available on the subject of ignition probability, most likely because such statistics are difficult to establish following an accident involving an ignited release. It may be noted that the extent of available data for the other critical element of the risk quantification, that is leak frequency, is quite a bit better although not perfect.
Further, there is very little experimental data available, due to the difficulty or impossibility of establishing realistic values through laboratory experiments. The type of data that are available is limited to flash points, auto ignition temperatures, etc.
Ignition probability models have been published in several textbooks and papers. These models reflect leak rates and module volumes, but seldom include anything approaching design and operation details. The collection of leak and ignition data by UK HSE is the most extensive online data collection scheme in existence.
There has been no ignited hydrocarbon leak with leak rate above 0.1 kg/s since 19 November 1992. In this incident a small gas leak, probably in the order of just above 0.1 kg/s, was ignited most likely by grinding sparks during modification work. Prior to that there was an oil leak from an export pump, where the leak source was also the ignition source, a failed seal on the export pump during normal...
