Modelling Software Failures of Digital I&C in Probabilistic Safety Analyses based on the TELEPERM (R) XS Operating Experience

Digital instrumentation and control (I&C) systems appear as upgrades in existing nuclear power plants (NPPs) and in new plant designs. In order to assess the impact of digital system failures, quantifiable reliability models are needed along with data for digital systems that are compatible with existing probabilistic safety assessments (PSA). The paper focuses on the modelling of software failures of digital I&C systems in probabilistic assessments. An analysis of software faults, failures and effects is presented to derive relevant failure modes of system and application software for the PSA. The estimations of software failure probabilities are based on an analysis of the operating experience of TELEPERM (R) XS (TXS). For the assessment of application software failures the analysis combines the use of the TXS operating experience at an application function level combined with conservative engineering judgments. Failure probabilities to actuate on demand and of spurious actuation of typical reactor protection application are estimated. Moreover, the paper gives guidelines for the modelling of software failures in the PSA. The strategy presented in this paper is generic and can be applied to different software platforms and their applications.