A benchmark for quantitative fault tree reliability analysis

摘要

Reliability analysis of critical computer-based systems is often performed using software tools, typically using fault tree analysis. Many software tools are available, either commercially or from university research groups, and each tool uses different solution techniques, ranging from fast approximations to more complex analysis such as Markov analysis or binary decision diagrams. Analysts are thus faced with the difficult task of validating a tool which is being considered for use in a real application. One approach to verification might be to select a set of representative examples, and compare the tool results with some known set of "correct" solutions. However, the development of a reliable benchmark can be time consuming, and the determination of the correct result may be very difficult. It would therefore be very useful if there was a published benchmark, with the correct solutions, from which the analyst could select test cases. Towards this goal the authors present a set of example systems and their analysis. They use these examples to help validate theirr dynamic fault tree analysis tool, Galileo. Their goal is to gather a set of representative examples which are reasonably challenging and which are derived from real systems.