Uncertainty assessment of reliability estimates for safety-instrumented systems

Abstract

Reliability estimates play a crucial role in decision making related to the design and operation of safety-instrumented systems. A safety-instrumented system is often a complex system whose performance is seldom fully understood. The safety-instrumented system reliability estimation is influenced by several simplifications and assumptions, both about the safety-instrumented system and its operating context, and therefore subject to uncertainty. If the decision makers are not aware of the level of uncertainty, they may misinterpret the results and select a safety-instrumented system design that is either too complex or too simple, or with an inadequate testing strategy, to provide the required risk reduction. This article elucidates the uncertainties related to safety-instrumented system reliability estimation. The article is limited to safety-instrumented systems that are operated in a low-demand mode, for which the probability of failure on demand is the standard reliability measure. The uncertainty of the probability of failure on demand estimate is classified as completeness uncertainty, model uncertainty, and parameter uncertainty and each category is thoroughly discussed. It is argued that the completeness uncertainty is the most important for safety-instrumented system reliability analyses, followed by parameter and model uncertainty. It is further argued that uncertainty assessment should be an integrated part of any safety-instrumented system reliability analysis, and that the analyst should communicate her judgment about the uncertainty to the decision-makers as part of the analysis results.