A novel approach for software reliability analysis of digital instrumentation and control systems in nuclear power plants

Abstract

This work presents a novel, integrated approach to the reliability analysis of digital instrumentation and control systems by incorporating Bayesian belief network (BBN), human reliability analysis (HRA), and common cause failure (CCF) modeling techniques. The Bayesian and HRA-Aided Method for the Reliability Analysis of Software (BAHAMAS) provides consideration of software development life cycle (SDLC) processes and their influence on software reliability. It is assumed that software failures can be traced to human errors in the SDLC, which can be modeled with HRA methods. Additionally, a system's reliability can be predicted based on how its SDLC quality compares with existing similar systems. A case study demonstrates the quantification of results from a hazard analysis of a digital reactor trip system. The case study shows agreement with values reported in the literature. BAHAMAS is shown to be a flexible tool whose application is designed to conveniently incorporate with conventional probability risk assessments.