(54l) An Approach to QRA Using Modeling to Develop Event Frequencies

The traditional QRA in the Oil & Gas industry relies on historical data to account for the frequency of the loss of containment. Such approaches tend to emphasize mitigation over prevention and do not take into account process/cultural/procedural contributors specific to each operation, which have important effects into the overall risk profile.

For over four (4) decades a variety of industries including, but not limited to: nuclear, aerospace, and DOE/DoD chemical facilities have used an approach to QRA that allows for modeling of systems and human actions that may contribute to loss of containment and therefore, emphasize “prevention” as well as “mitigation.” This approach still relies on historical data to develop and quantify the model for loss of containment. This is done for a variety of reasons that include:

1. Allows for modeling of complex, as well as simple, systems with significant dependencies.
2. Allows for modeling of human interface for actions that are intended to prevent or mitigate the accident, as well as those that may initiate and/or accelerate accident progression.
3. Provides insights into “prevention” as well as “mitigation” of risk.

In this approach, the accident is studied in two phases:

• Pre-loss of containment – Involves definition of the initiating events leading to the loss of containment, including “accident escalation” (a.k.a. “accident progression”) scenarios. The initiating event could be a disturbance to the system which could include any deviation from the normal safe operating envelope. The disturbance may be caused by internal (pump failure, mal-operation of equipment, etc.) or external factors (extreme weather, seismic, etc.). Depending on the complexity of the system, the ramifications of the pre-loss of containment scenarios can be modeled using fault trees or simpler tools such as FMEAs. The complexity of the systems is usually assessed based on the number of components involved in the containment system, the number of functional dependencies linking these components, and the human interface (i.e., the increased system complexity in terms of the inter-dependencies or human interface).
  
• Post-loss of containment - The final consequence following the progression of each event derived from the initial loss of containment is modeled using an event tree. The frequency of each consequence is calculated by multiplying the frequency of the initial loss of containment by each outcome frequency.

This approach not only provides estimates and insights into risk, it is also a powerful tool in improving operability and protecting assets through better design, operation and preventive maintenance. This has been demonstrated in many industries instituting formal risk management practices that have been accepted and encouraged by authorities having jurisdiction.

This paper discusses the application of the method in the oil and gas industry and chemical processing. The oil and gas application is assessing production platform safety and optimizing levels of hydrocarbon production. The chemical processing application is assessing risk associated with release of flammable refrigerants to provide insights for safe application of these refrigerants through the entire life cycle, from design to decommissioning.