(65bd) The Application of Frequency Data in Quantitative Risk Assessment

Quantitative Risk Assessment (QRA) is a vital resource for estimating risk in the process industry. There are four primary steps involved in conducting a QRA: hazard identification, frequency analysis, consequence analysis, and risk calculation (the combination of frequency and consequence to determine the risk). Frequency analysis is a key component of the QRA and can have a major impact on the risk profile.

The primary method of estimating event frequency for QRA utilizes historical leak frequency data, of which the Hydrocarbon Release Database (HCRD) is the most widely used leak frequency database. One main issue with applying historical frequency is not having transparency of the causes of the events. Using the historical data approach, a single value is generated for the frequency of an event occurring, with no association to the cause (e.g. corrosion, dropped object, etc.). It has also become widely accepted to apply these offshore databases to onshore processes. There are many concerns in applying these databases to onshore processes, as there are differences in the types of equipment, the hazards the equipment is exposed to, and the safety management systems in place.

When QRAs were first conducted during the 1980s, they were an exhaustive and detailed process. In today’s fast track industry, often frequency analysis is conducted without undertaking an evaluation to determine if the frequency being applied is appropriate for the hazard at hand. Leak frequency analysis has become an isolated activity, rather than relying on frequencies established during HAZOP and reliability assessments such as fault tree analysis.

An alternative method to employing historical databases to estimate event frequency is conducting fault tree analysis. Fault tree analysis involves dissecting a hazard into its causes, using a “top down” approach, beginning with hazard identification and working down to a level in which applicable frequency data is available. Fault trees provide a logical relationship between the hazard and its causes.

If historical leak frequency data is employed, which is independent of the cause, the resulting risk is also independent of the cause. Therefore, mitigation measures and the decision making process tend to focus on the back-end consequence of the event, rather than the cause of the event.

Fault tree analysis, although it has had limited application in onshore and offshore risk assessment, provides a major advantage of providing a link between the cause of an event and the associated frequency. This transparency between the event frequency and cause enables us to identify major contributors to the risk and provide risk reduction measures based on what is actually causing the event to occur.

The objective of this paper is to provide an alternative to employing offshore historical frequency data to QRA and demonstrate that the alternative method can provide major advantages in mitigations by linking the estimated risk to the cause and focusing the decision making process on the front-end of the problem rather than solely on the back-end.