Risk-Based Approach – Frequency Analysis

The primary purpose of this paper is to provide tools, guidance and criteria for finding and appropriately using failure rate data needed to perform a risk-based quantitative analysis, as it is critical to understanding of failure rates, their origin and limitations.

A risk-based quantitative analysis requires the estimation of two key variables: the frequency that an event will occur and the consequence, which is the logical and expected impact of that event. Multiple methodologies and approaches are available to reasonably predict the consequences of a chemical release, fire and/or explosion on manufacturing equipment, people and the environment. The technology to do so is well developed and is enhanced when new information becomes available and more powerful computational tools evolve. However, frequencies and probabilities of enabling events are more difficult to predict and criteria must be established and followed from sources such as historical data, experiments and expert opinion. The primary purpose of this paper is to provide tools, guidance and criteria for finding and appropriately using failure rate data needed to perform a risk-based quantitative analysis (see Figure 01) as it is critical to understand failure rates, their origin and limitations.

The estimation of the likelihood of all Loss of Containments scenarios (LOCs) identified during the hazard identification phase is the main topic of this paper. The frequency analysis can be conducted using historical data, specific plant data (if available), using international references for generic process equipment failure rates and developing detailed fault trees for defining specific LOCs.

Generic failure data: Failure rate data generated from information collected on plant equipment failures are referred to as plant-specific data. Plant-specific data reflect the plant's processes, environment, maintenance practices and choice and operation of equipment. Data accumulated from a variety of plants and industries, such as nuclear power plants, Chemical Process Industry (CPI) or offshore petroleum platforms, is called generic data. With data from many sources, generic failure rate data can provide a much larger data set.