Red Room - Session 9

2:15 to 3:15 p.m. Wednesday April 22, 2015

Uncertainty in RAM Analysis

The outcome of a Reliability, Availability and Maintainability (RAM) analysis is often presented as a fixed number (e.g., expected system availability). However, in reality this number is far from "fixed." The reliability data of individual components used to carry out these analyses is based on populations of components or on expert judgment, which means that the point value that is used has an uncertainty. As a result, the outcome of the RAM analysis is not a fixed number but a probability density function.

This presentation describes the incorporation of information on the confidence (or uncertainty) of input data into the failure model, using Isograph's Reliability Workbench® software. By incorporating the confidence information of the input data, the simulation results also contain a variance that can be presented using a confidence interval. This provides information on the uncertainty of the RAM analysis results, which can be very valuable to asset owners. The uncertainty in the performance level of an asset is essential for effective asset and risk management based on ISO 55000.

Key Words: RAM, Fault Tree, Uncertainty, Confidence Interval, Failure Data, Asset Management

George de Schutter and Peter van Steenderen

Royal HaskoningDHV

The Netherlands