Modeling dynamic scenarios for safety, reliability, availability, and maintainability analysis

Abstract

Safety analysis uses probability combinatorial models like fault tree and/or event tree. Such methods have static basic events and do not consider complex scenarios of dynamic reliability, leading to conservative results. Reliability, availability, and maintainability (RAM) analysis using reliability block diagram (RBD) experience the same limitations. Continuous Markov chains model dynamic reliability scenarios but suffer from other limitations like states explosion and restriction of exponential life distribution only. Markov Regenerative Stochastic Petri Nets oblige complex mathematical formalism and still subject to state explosions for large systems. In the design of complex systems, distinct teams make safety and RAM analyses, each one adopting tools better fitting their own needs. Teams using different tools turns obscure the detection of problems and their correction is even harder. This work aims to improve design quality, reduce design conservatism, and ensure consistency by proposing a single and powerful tool to perform any probabilistic analysis. The suggested tool is the Stochastic Colored class of Petri Nets, which supplies hierarchical organization, a set of options for life distributions, dynamic reliability scenarios and simple and easy construction for large systems. This work also proposes more quality rules to assure model consistency. Such method for probabilistic analysis may have the effect of shifting systems design from “redundancy, segregation and independency” approach to “maintainability, maintenance and contingency procedures” approach. By modeling complex human and automated interventional scenarios, this method reduces capital costs and keeps safety and availability of systems.