Abstract
Many of the nation’s large-scale physical infrastructure systems are commonly composed of interconnected and intra- and interdependent subsystems, which in their essence constitute systems of systems (SoS) with multiple functions, operations, and stakeholders. Their complexity is characterized by the highly interconnected and interdependent physical, cyber, organizational and economic subsystems through shared resources, decisions and states, which constitute a major source of systemic risks inherent to the system and pose great challenges in their risk modeling, assessment, and management. To meet the increasing needs of reliable services provided by these infrastructure systems, system owners and decision makers need tools to foresee potential emergent forced changes from within and outside the system and to understand their impacts so that efficient risk management strategies can be developed.
Risk analysis of complex SoS requires a systemic and holistic approach that integrates multiple perspectives, models and tools. The focus of this dissertation is to develop a systemic framework of precursor analysis, which supports the design of an effective and efficient precursor monitoring system having the ability to i) identify indicators or warnings of dynamic and evolving risks to system failure; (ii) monitor critical precursors to system failure through continuously tracking and observing triggering changes in the states of the system; and (iii) reduce the hindsight bias frequently observed between pre- and post- accident risk assessment when using precursors. This pro-active and dynamic anticipatory analysis is supported by meta-modeling the functional components and subsystems of the SoS, and their relationships in a control structure and is achieved through a process of precursor identification, prioritization, detection, and evaluation.
The identification of precursors to system failure requires an understanding of system failure mechanism. This dissertation explores potential sources of systemic risks in complex SoS through analyzing a unique failure mode of the system in a nonlinear dynamic multi-objective decision process. It demonstrates that the decision maker’s inappropriate preference among multiple competing objectives and the interdependencies between uncoordinated subsystems contribute to the failure of complex SoS even though all its components are functioning correctly. The results also suggest that an optimal decision strategy doesn’t necessarily guarantee system safety. Through quantifying the level of subsystem interdependency caused by common states, this dissertation develops a method to decompose interconnected subsystems within SoS and a method to coordinate multiple subsystems in a decentralized way.
This dissertation demonstrates the theories and methodologies with a case study on the US highway bridge system. Highway bridges, which constitute large- and multi-scale physical infrastructure systems, and are essential elements of transportation networks, have a large number of interconnected and interdependent sub-systems, with broad social and economic consequences from bridge failure. The precursor analysis framework allows examining the impacts of current bridge inspection, maintenance, and decision practices on the overall reliability of bridge infrastructure systems; enables decision makers to make more timely and informed decisions to efficiently allocate limited risk management resources; and thus, prevent future severe consequences resulting from future bridge failures.
