Dynamic System Reliability

Modeling and Analysis of Dynamic and Dependent Behaviors

Liudong Xing, Gregory Levitin, Chaonan Wang

• Publisher: Wiley
• Year: 2019
• Print ISBN: 9781119507635
• eBook ISBN: 9781119507680
• Edition: 1
• Topic: Technology & Engineering
• Subtopic: Engineering General
• Index: Technology & Engineering

Offers timely and comprehensive coverage of dynamic system reliability theory

This book focuses on hot issues of dynamic system reliability, systematically introducing the reliability modeling and analysis methods for systems with imperfect fault coverage, systems with function dependence, systems subject to deterministic or probabilistic common-cause failures, systems subject to deterministic or probabilistic competing failures, and dynamic standby sparing systems. It presents recent developments of such extensions involving reliability modelling theory, reliability evaluation methods, and features numerous case studies based on real-world examples. The presented dynamic reliability theory can enable a more accurate representation of actual complex system behavior, thus more effectively guiding the reliable design of real-world critical systems.

Dynamic System Reliability: Modelling and Analysis of Dynamic and Dependent Behaviors begins by describing the evolution from the traditional static reliability theory to the dynamic system reliability theory, and provides a detailed investigation of dynamic and dependent behaviors in subsequent chapters. Although written for those with a background in basic probability theory and stochastic processes, the book includes a chapter reviewing the fundamentals that readers need to know in order to understand contents of other chapters which cover advanced topics in reliability theory and case studies.

• The first book systematically focusing on dynamic system reliability modelling and analysis theory
• Provides a comprehensive treatment on imperfect fault coverage (single-level/multi-level or modular), function dependence, common cause failures (deterministic and probabilistic), competing failures (deterministic and probabilistic), and dynamic standby sparing
• Includes abundant illustrative examples and case studies based on real-world systems
• Covers recent advances in combinatorial models and algorithms for dynamic system reliability analysis
• Offers a rich set of references, providing helpful resources for readers to pursue further research and study of the topics

Dynamic System Reliability: Modelling and Analysis of Dynamic and Dependent Behaviors is an excellent book for undergraduate and graduate students, and engineers and researchers in reliability and related disciplines.

1
Introduction

The advances and interdisciplinary integration of science and technology are making modern engineering and computing systems more and more complex. For modern systems (especially those in, e.g. wireless sensor networks, Internet of Things (IoT), smart power systems, space explorations, and cloud computing industries), dynamic behavior and dependence are typical characteristics of the systems or products. System load, operating conditions, stress levels, redundancy levels, and other operating environment parameters are variables of time, causing dynamic failure behavior of the system components as well as dynamic system reliability requirements. In addition, components of these systems often have significant interactions or dependencies in time or functions. Effects of these dynamic and dependent behaviors must be addressed for accurate system reliability modeling and analysis, which is crucial for verifying whether a system satisfies desired reliability requirements and for determining optimal design and operation policies balancing different system parameters like cost and reliability. As a result, reliability modeling and analysis of modern dynamic systems become more challenging than ever.

Traditional reliability modeling methods, such as reliability block diagram [1] and fault tree analysis [2], can define the static logical structure of the system, but they lack the ability to describe dynamic state transfers of the system, and component fault dependencies and propagations. It is difficult or impossible to accurately reflect the actual behavior of modern complex fault‐tolerant systems using the traditional reliability models. In other words, failure to address effects of dynamic behavior and dependencies of modem systems makes the reliability analysis results obtained using the traditional reliability models far from the actual system reliability performance, misleading the system design, operation, and maintenance efforts.

Different from the traditional static reliability modeling, the dynamic reliability theory considers that a system failure depends not only on the static logical combination of basic component failure events, but also on the timing of the occurrence of the events, correlations or interrelationship of the events, and impacts of operating environments. Therefore, the dynamic system reliability theory can provide a more accurate representation of actual complex system behavior, more effectively guiding the reliable design of real‐world critical systems. The dynamic system reliability theory is the evolution and improvement of the traditional reliability modeling theory, and its research will promote the development and application of complex systems engineering.

This book focuses on dynamic reliability modeling of fault‐tolerant systems with imperfect fault coverage, functional dependence, deterministic or probabilistic common‐cause failures, deterministic or probabilistic competing failures, as well as standby sparing.

Specifically, imperfect fault coverage is an inherent behavior of fault‐tolerant systems designed with redundancies and automatic system recovery or reconfiguration mechanisms [3–5]. Just like any system component, the system recovery mechanisms involving fault detection, fault location, fault isolation, and fault recovery will likely not be perfect; they can fail such that the system cannot adequately detect, locate, isolate, or recover from a fault occurring in the system. The uncovered component fault may propagate through the system, causing an extensive damage to the system, sometimes failure of the entire system. Further, it is observed that the extent of the damage from an uncovered component fault occurring in a system with the hierarchical nature may exhibit multiple levels due to the layered recovery [6]. The traditional imperfect fault coverage concept has been extended to the modular imperfect fault coverage to model multiple levels of uncovered failure modes for components in hierarchical systems [7].

Functional dependence occurs in systems where the failure of one component (or, in general, the occurrence of a certain trigger event) causes other components (referred to as dependent components) within the same system to become unusable or inaccessible. A classic example is a computer network where computers can access the Internet through routers [8]. If the router fails, all computers connected to the router become inaccessible. It is said that these computers have functional dependence on the router.

In the case of systems with perfect fault coverage, the functional dependence behavior can be addressed as logic OR relationship [9]. However, for systems with imperfect fault coverage, the logic OR replacement method can lead to overestimation of system unreliability because it allows the disconnected dependent components (in the case of the trigger event occurring) to contribute to the system uncovered failure probability if they can fail uncovered. However, since these dependent components were disconnected or isolated, they could really not generate propagation effect or bring the system down [10]. New algorithms are required for addressing the coupled functional dependence and imperfect fault coverage behavior.

In addition to the imperfect fault coverage, common‐cause failures are another class of behavior that can contribute significantly to the overall system unreliability [11–13]. Common‐cause failures are defined as “A subset of dependent events in which two or more component fault states exist at the same time, or in a short time interval, and are direct results of a shared cause” [11] . Most of the traditional common‐cause failure models assumed the deterministic failure of the multiple components affected by the shared root cause. Recent studies extended the concept to model probabilistic common‐cause failures, where the occurrence of a root cause results in failures of multiple system components with different probabilities [14–16].

As one type of common‐cause failures, a propagated failure with global effect (PFGE) originating from a system component can cause the failure of the entire system [17]. Such a failure can occur due to the imperfect fault coverage or destructive effect of a component failure on other system components (like overheating, explosion, etc.). However, PFGE may not always cause the overall system failure in systems with functional dependence behavior. Specifically, if the trigger event occurs before PFGEs of all the dependent components, these PFGEs can be isolated deterministically and thus cannot affect other parts of the system. On the other hand, if PFGE of any dependent component occurs before the trigger event, the failure propagation effect takes place, crashing the entire system. Therefore, there exist competitions in the time domain between the failure isolation and failure propagation effects, causing distinct system statuses [18,19].

The pioneering works on addressing such competing failures in system with functional dependence have focused on deterministic competing failures, where the occurrence of the trigger event, as long as it happens first, can cause deterministic or certain isolation effect to any failures originating from the corresponding dependent components. Recent studies [20,21] have revealed that in some real‐world systems, e.g. systems involving relayed wireless communications, the failure isolation effect can be probabilistic or uncertain. Consider a specific example of a relay‐assisted wireless sensor network where some sensors pr...

Table of contents

• Cover
• Table of Contents
• Foreword by Dr. Andre Kleyner, Series Editor
• Preface
• Nomenclature
• 1 Introduction
• 2 Fundamental Reliability Theory
• 3 Imperfect Fault Coverage
• 4 Modular Imperfect Coverage
• 5 Functional Dependence
• 6 Deterministic Common‐Cause Failure
• 7 Probabilistic Common‐Cause Failure
• 8 Deterministic Competing Failure
• 9 Probabilistic Competing Failure
• 10 Dynamic Standby Sparing
• Index
• End User License Agreement