Foundations of the UPM common cause method

摘要

Common cause failures (CCF) have a significant impact on the risk of complex engineering systems, such as nuclear plants. Consequently, modelling this class of events is an important part of risk assessments, and the authorities that are responsible for the safe and efficient functioning of the aforementioned systems have taken this issue on hoard. The current approach in the UK towards CCF modelling is the Unified Partial Method (UPM), which is based on the Partial Beta Factor Model and the Cut-Off Methodology. The UPM scoring system itself is based on an additive weighting factors approach. This is closely allied to the additive value function approach of Multialtribute Value Theory. However, within MAVT there are many other possible models. We explore whether the preferential independence axiom, required for an additive value function, is conceptually appropriate. In addition we discuss the impact of new data availability, in particular from the ICDE project. The research discussed here is part of an on-going research project, funded by the Health and Safety Executive.