Text

theses

The continuous improvements in systems engineering and the unprecedented rate of technological advances not only take the quality and reliability engineering to the forefront, but also bring the large and complex engineered systems into practical use. On the one hand, the ever-rising expectations of the customers of the reliability of products and services have enhanced the design, operation and maintenance phases during their life cycles. Moreover, cascading effects, significant damages and interruptions of services caused by failures of large and complex systems, such as telecommunication networks, power grids, transportation systems, healthcare delivery systems, information systems, financial systems and supply chain systems, have aroused researchers' attention.

The last two decades have witnessed increasing reliance of these systems on computers, sensors, software and applications that have become targets of cyber attack and software failures with major consequences. Natural disasters and hazards such as floods, hurricanes and earthquakes particularly cause significant disruptions of the systems’ services. Restorations of their functionality under limited resources and time constraints have given rise to the assessment of such systems’ resilience. However, traditional reliability metrics are inadequate to assess the resilience characteristics in many applications and critical infrastructure sectors. Therefore, resilience as a new extension of reliability metrics has been gradually and widely used to evaluate the performance of large and complex systems.

Ideally, system recovery is “optimized” when all failed (and degraded) units are recovered immediately after the hazard; which is unrealistic due to the limited recovery resources and repair times needed to restore the system to its operational levels. Therefore, to recover system performance to a desired level within the shortest period, it becomes important to determine the sequence in which failed and degraded units are repaired sequentially (or simultaneously when possible). Specifically, it is necessary to obtain the criticality of the failed and degraded units during the recovery process and allocate the repair resources to the most important units which have the highest impact on the system recovery by using an importance measure (IM). IM is also used for identifying system design weakness and component (or subsystem) failures that are crucial to the system performance, and therefore determine the allocation of redundancy or repair resources to achieve system performance improvement.

In this thesis, we provide a detailed overview of potential hazards and methods of their predictions and quantification; we present several definitions of resilience as well as methods of its assessment in different applications; we also present the development of importance measures and compare them in different scenarios; we review cascading failure occurring in systems and models of their assessment and prevention. We propose general resilience metrics for non-repairable and repairable systems and demonstrate their estimation through applications. We finally propose approaches to prioritize units of the system in order of their importance to the system functions and to optimize the maintenance resources in order to recover system performance to a desired level within the shortest period.

ETD_9446

M.S.

Includes bibliographical references

by Xi Chen

doi:10.7282/t3-69za-jr69

ETD graduate

The author owns the copyright to this work.