Ola Bäckström, Jan-Erik Holmberg, Mariana Jockenhövel-Barttfeld, Markus Porthin, Andre Taurines,
A project is ongoing, financed by Nordic nuclear safety research (NKS), The Finnish Research Programme on Nuclear Power Plant Safety (SAFIR2014) and Nordic PSA group (NPSAG), with the intent to provide guidelines to analyse and model digital systems in probabilistic safety assessment (PSA), using traditional reliability analysis methods (FMEA, Fault tree analysis).
This report discusses software reliability in this context. The report proposes a method for the evaluation and quantification of reactor protection system (RPS) software failures. The proposed method will use operational history to estimate the fatal failure probability within system software (operating system, runtime), and use an indirect method for the estimation of failure probability within application software (non-fatal and fatal failures). The quantification for application software is based on two main measures, complexity and the degree of verification and validation of the software.
Collection of data and its challenges will also be discussed. Some data collected for a software platform will be discussed, and used as an example of the difficultness — and challenge — to collect data.
PSA, Software reliability, Operational history data