Testing the Linearity Assumption in Leak Frequencies for Offshore Process Equipment

Offshore QRAs normally assume that an installation's annual leak frequency depends linearly on its equipment count. This assumption is embedded in all generic leak frequency data for specific equipment items. To our knowledge, it has never been validated.
In this paper, we examine whether the relationship between leak frequency and the number of equipment items on an installation really is linear. Our chosen dataset is the UK offshore hydrocarbon release database. By accessing the original equipment population data, we have been able to calculate leak frequencies for individual installations varying widely in total equipment counts. Our results show significant deviations from the linear model.
After considering various possible interpretations, we conclude that lower frequencies are typically observed on small normally-unattended installations (NUIs), and also on large installations which have more opportunity to learn from experience and more incentive to invest in reliability. Excluding NUIs, there is a consistent decline in the leak frequency per equipment item as installation size increases.

Keywords: Risk assessment; leak frequencies; process equipment.