The treatment of dependent failures is one of the most controversial subjects in reliability and risk analyses. The difficulties are specially underlined in the case of highly redundant systems, when the number of redundant components or trains exceeds four. The Common Load Model (CLM), originally defined in the 70’ies, differs from other CCF models, as it relies on a specific physical analogy. The model was extended in late 80’ies with specific aims to model highly redundant systems. Practical use of the model requires a computer tool, but in highly redundant systems the benefits give an evident payback.
The extended method ECLM details have not been publicly available before. However, it has been developed with certain funding via the Nordic PSA group (NPSAG) including SSM. It has therefore been an interest to have the method properly documented and published. The objective with this report is to describe the mathematical details in the definition of Extended CLM (ECLM), which is developed to better suit for modelling of highly redundant systems. ECLM has four parameters, while the initial CLM was a two-parameter model. This report focuses on parameter definition and interface between a dedicated CCF analysis and PSA models. The estimation of model parameters is described for Maximum Likelihood and Bayesian approaches. Also a comparison is presented with respect to other common CCF models. Special extensions are described for time-dependent modelling and asymmetric CCF groups.
The result is that the method is documented and available for the stakeholders, including SSM. The report is also published as an NPSAG report.