Quantification of reactor protection system software reliability based on indirect and direct evidence

This paper presents a method for the quantification of software failures in a reactor protection system in the context of probabilistic safety assessment (PSA) for a nuclear power plant. The emphasis of the method is on the quantification of the failure probability of an application software module, which can lead to the functional failure modes: failure to actuate on demand a specific instrumentation and control (I&C) function or spurious actuation of a specific I&C function. The quantification is based on two main metrics, complexity of the application software and the degree of verification and validation of the software. The relevance of common cause failures and an analysis of the impact of fatal and non-fatal failures on the system will be covered by the discussion. Collection of operational data and challenges to use it for software reliability quantification will also be discussed. The outlined quantification method offers a practical and justifiable approach to account for software failures that are usually ignored in current PSAs.