Integrating Software into PRA (PROBABILISTIC RISK ANALYSIS)

Abstract

Probabilistic Risk Assessment (PRA) is a methodology consisting of techniques to assess the probability of failure or success of a system. In many modern technological systems, especially safety critical systems such as space systems, nuclear power plants, medical devices, defense systems, etc, PRA has been proven to be a systematic, logical, and comprehensive methodology for risk assessment, for the purpose of increasing safety in design, operation and upgrade, and for reducing the costs in design, manufacturing, assembly and operation.

Software plays an increasing role in modern safety critical systems. A significant number of failures can be attributed to software failures such as the well-known Therac-25 radiation overdose accidents, the Mars Climate Orbiter, Mariner I Venus Probe and Ariane 5 accidents. Unfortunately current PRA practice ignores the contributions of software due to a lack of understanding of the software failure phenomena. The objective of our research is to develop a methodology to account for the impact of software on system failure that can be used in the classical PRA analysis process.

To develop the methodology, a systematic integration approach is studied and defined. Next, a taxonomy of software-related failure modes is established and validated. The software representation in fault trees and event trees is defined. A test-based approach for modeling and quantifying the software contribution is presented. A Case study is provided to validate the framework.

This study is the first systematic effort to integrate software risk contributions into PRA.