Probabilistic modelling of common cause failures in digital I&C systems - Literature review

This report presents a state of the art review on probabilistic risk assessment of CCFs in digital instrumentation and control (I&C) systems of nuclear power plants. It covers a literature study and a questionnaire to Finnish nuclear power companies. Both software and hardware CCFs are in the scope of the report. There is relatively little literature addressing these CCFs. Concerning both software and hardware, lack of data is the main challenge, and there is need for data collection and method development activities.

Software CCF probabilities are usually based on either expert judgments or operating experience. Software reliability analysis methods in scientific literature do not usually address CCFs specifically. Only one method that focuses on software CCFs of a digital reactor protection system was found.

It is generally agreed that CCFs between identical redundant software modules can be modelled assuming full dependency. However, when there is some diversity present, the modelling is challenging. Some guidance exists for such cases nevertheless.

Hardware CCFs can be analysed according to normal CCF analysis principles. However, lack of data on digital I&C components often makes it necessary to use generic parameters or engineering judgment -based methods, which can lead to quite conservative results. Large and asymmetric CCF groups are one particularly challenging area related to digital I&C hardware.