System Reliability in Quantitative Risk Assessment

This course presents the basic concepts of reliability analysis of complex systems, one of the corner stones of risk assessment and management of technological systems. It starts with a description of the fundamental procedural steps of a quantified risk assessment and identification of the scientific disciplines and available methodological tools to complete each step.

A system (scenario-based) approach is employed, where the source of risk is considered as a system with constituent components or elements. Risk is then defined and determined as a set of triplets {si, pi, ci}: si are the scenarios (sequence of events) that lead to the undesired consequences, ci, and pi are the probabilities with which these scenarios are expected to occur.

The course focuses on methodologies and corresponding tools that can provide the scenarios (si) and the corresponding probabilities of occurrence (pi). To this end, the complex system is modeled as a collection of interconnected subsystems or components among which there are functional interdependencies. Methods for modeling the functional interdependencies of the components will be presented both for two-state as well as for multi-state components. These methods model the states of the complex system as collection of states of its components. Next, the principles of probability are employed to assess the probability of the various system states in terms of the probability of the component states.

Fundamental concepts and measures of reliability will be introduced at a components level. Then complex systems (multi-component) will be addressed.

Static or on demand events will be examined. Methods to be studies include block-diagrams, fault-trees, event-trees. Techniques for quantification of these models including dependence analysis will be presented. Importance analysis will be examined providing methods for determining the relative contribution of each component into the proper function of a complex system, as well as, its contribution to the probability of system failure. The use of data (observed behavior of the system and its components) to derive the values of various reliability parameters will be presented.

Throughout the course, practical examples of systems from the chemical, nuclear, electricity generation, water, energy and railway industry, as well as, examples from occupational risk and land use domain will be used to demonstrate the various methods and techniques.