There are several pieces to risk assessment and risk management that allow facility operators to evaluate and measure what is actually occurring in their facilities, including the review of historical event data to identify what went wrong and for near misses, what almost happened. This paper will describe techniques that have been used in the commercial nuclear power and other industries including NASA, FAA, and US Department of Energy production facilities to collect and apply historical experience data to improve safety and operability.
As described by EnergyWire, “Partial failures happen regularly on oil rigs, with a worker catching the errors and correcting them before a major blowout. These are not always reported to regulators, but they are logged carefully and stored as raw data on rigs and onshore in Houston.”
One data analysis approach using accident sequence precursor (ASP) analysis uses this type of raw incident data and risk analysis techniques to help identify potential accidents before they occur. Reports of plant incidents are reviewed to identify those potentially involving failures in systems that provide protective functions against severe accident-related initiating events or the initiating events themselves. Those events selected undergo one- or two-engineer review(s) to determine if the reported event should be examined in greater detail. Events determined to be potentially significant as a result of this initial review are then subjected to a thorough, detailed analysis. This extensive analysis is intended to identify those events considered to be precursors to potential severe accidents, either because of an initiating event or because of failures that could have affected the course of postulated off-normal events or accidents. A variety of other data sources about the same event are also reviewed and compared to thoroughly understand the contributors to each event.
Quantification of the risk significance of an accident sequence precursor involves the determination of a conditional probability of a subsequent severe accident given the failures observed during an operational event. This probability is estimated by mapping the observed failures onto the ASP accident sequence models (event trees and linked fault trees modified to reflect the event), which depict potential paths to the severe accident, and by calculating a conditional probability of the severe accident.
The ASP approach is similar to LOPA or Bow Tie Analysis in that it postulates accident scenarios and the paths that could lead to a potential accident, but differs in that it uses insights from actual historical near-miss event data to identify the branch points which, if different, could have led to a severe accident.
The information gained from data analysis such as ASP not only provides quantitative data, but identifies the types of events that should be included in risk models. ASP can also facilitate the prioritization or ranking of precursor categories based on both frequency of occurrence and risk significance. In addition, ASP can help identify events that reflect unusual failure modes with the potential to compromise continued safety functions.