Extensions of Inoperability Input-Output Modeling for preparedness decisionmaking: Uncertainty and inventory

Models are built to answer specific questions. The Inoperability Input-Output Model (IIM), the cornerstone of the work presented in this dissertation, and its derivatives were developed to answer the following question: what are the impacts (e.g., proportion of inoperability and dysfunctionality, economic loss) on the sectors of an interdependent system of infrastructures given a disruptive event affecting multiple sectors? However, to answer such specific questions, models must have the ability to address many aspects of the system. The quality and effectiveness of the IIM and its derivatives, as a modeling enterprise, are subject to model assumptions, uncertainty in model parameters, and topology, among others, which may hinder the complexity required to realistically deal with the specific questions that we seek to answer. Short of modifying the basic, widely-accepted Leontief-based inoperability model for which a considerable data collection effort is undertaken annually, this dissertation attempts to compensate for the several perceived deficiencies in the IIM and its derivatives to better quantify the efficacy of preparedness risk management strategies for interdependent infrastructure and economic sectors.; The modeling capability of the IIM, and namely the Dynamic Inoperability Input-Output Model (DIIM), is enhanced with the inclusion of uncertainty analysis. Uncertainty in capturing the behavior of disruptive events and their effect on IIM and DIIM parameters is addressed with probability distributions, not point estimates, in place of scenario-specific parameters. Uncertainty in the interactions among interdependent sectors is addressed with the integration of the DIIM with the Uncertainty Sensitivity Index Method (USIM).; Expected and conditional expected values of risk describing extreme and catastrophic events are highly dependent upon the probability distributions from which expected and conditional expected values are calculated. As parametric uncertainties and estimation errors in probability distributions can result in adverse effects in the understanding of extreme events, the sensitivity of the parameters of probability distributions when measuring typical events and extreme events is calculated discussed.; The DIIM is further extended to allow for preparedness options that include inventory policies as a means to delay inoperability following a disruptive event. Current IIM and DIIM approaches are unable to model such inventory policies.