More Like this

1. Quantifying the Interdependency Strength Across Critical Infrastructure Systems Using A Dynamic Network Flow Redistribution Model.

   Wang, Y.; Yu, J.; Baroud, H. (July 2020, Proceedings of the 30th European Safety and Reliability Conference and 15th Probabilistic Safety Assessment and Management Conference.)

   Baraldi, P.; null; Zio, E. (Ed.)

   Critical infrastructure networks are becoming increasingly interdependent which adversely impacts their performance through the cascading effect of initial failures. Failing to account for these complex interactions could lead to an underestimation of the vulnerability of interdependent critical infrastructure (ICI). The goal of this research is to assess how important interdependent links are by evaluating the interdependency strength using a dynamic network flow redistribution model which accounts for the dynamic and uncertain aspects of interdependencies. Specifically, a vulnerability analysis is performed considering two scenarios, one with interdependent links and the other without interdependent links. The initial failure is set to be the same under both scenarios. Cascading failure is modeled through a flow redistribution until the entire system reaches a stable state in which cascading failure no longer occurs. The unmet demand of the networks at the stable state over the initial demand is defined as the vulnerability. The difference between the vulnerability of each network under these two scenarios is used as the metric to quantify interdependency strength. A case study of a real power-water-gas system subject to earthquake risk is conducted to illustrate the proposed method. Uncertainty is incorporated by considering failure probability using Monte Carlo simulation. By varying the location and magnitude of earthquake disruptions, we show that interdependency strength is determined not only by the topology and flow of ICIs but also the characteristics of the disruptions. This compound system-disruption effect on interdependency strength can inform the design, assessment, and restoration of ICIs. 

   more » « less
2. GLOBAL SENSITIVITY ANALYSIS MEASURES BASED ON STATISTICAL DISTANCES

   https://doi.org/10.1615/Int.J.UncertaintyQuantification.2021035424  

   Nandi, Souransu; Singh, Tarunraj (October 2021, International Journal for Uncertainty Quantification)

   null (Ed.)

   Global sensitivity analysis aims at quantifying and ranking the relative contribution of all the uncertain inputs of a mathematical model that impact the uncertainty in the output of that model, for any input-output mapping. Motivated by the limitations of the well-established Sobol' indices which are variance-based, there has been an interest in the development of non-moment-based global sensitivity metrics. This paper presents two complementary classes of metrics (one of which is a generalization of an already existing metric in the literature) which are based on the statistical distances between probability distributions rather than statistical moments. To alleviate the large computational cost associated with Monte Carlo sampling of the input-output model to estimate probability distributions, polynomial chaos surrogate models are proposed to be used. The surrogate models in conjunction with sparse quadrature-based rules, such as conjugate unscented transforms, permit efficient calculation of the proposed global sensitivity measures. Three benchmark sensitivity analysis examples are used to illustrate the proposed approach. 

   more » « less
3. Ranking the Impact of Interdependencies on Power System Resilience using Stratified Sampling of Utility Data

   https://doi.org/10.1109/TPWRS.2023.3260119  

   Kelly-Gorham, Molly Rose; Hines, Paul D.H.; Dobson, Ian (January 2023, IEEE Transactions on Power Systems)

   It is well known that interdependence between electric power systems and other infrastructures can impact energy reliability and resilience, but it is less clear which particular interactions have the most impact. There is a need for methods that can rank the relative importance of these interdependencies. This paper describes a new tool for measuring resilience and ranking interactions. This tool, known as Computing Resilience of Infrastructure Simulation Platform (CRISP), samples from historical utility data to avoid many of the assumptions required for simulation-based approaches to resilience quantification. This paper applies CRISP to rank the relative importance of four types of interdependence (natural gas supply, communication systems, nuclear generation recovery, and a generic restoration delay) in two test cases: the IEEE 39-bus test case and a 6394-bus model of the New England/New York power grid. The results confirm industry studies suggesting that a loss of the natural gas system is the most severe specific interdependence faced by this region. 

   more » « less
4. An Exact Multiobjective Optimization Approach for Evaluating Water Distribution Infrastructure Criticality and Geospatial Interdependence

   https://doi.org/10.1029/2018WR024063  

   Abdel‐Mottaleb, N.; Ghasemi Saghand, P.; Charkhgard, H.; Zhang, Q. (July 2019, Water Resources Research)

   Abstract Failures within water distribution systems are usually not isolated and tend to propagate to corresponding transportation infrastructure, yet most criticality and resilience analyses of water distribution networks are conducted for the individual water infrastructure without accounting for interdependence. To address this research gap, this study investigates how the critical components identified within water distribution systems may be different when accounting for failure propagation to the transportation road network. In this study, failure propagation is assumed to be based on geospatial interdependence and unidirectional, starting from water distribution network components to transportation network components. A logical interaction network is constructed considering the interdependence between both infrastructures, and multiobjective optimization is used to solve for the critical water distribution components considering: quantity of failures, performance loss, and financial costs. This work presents a modular workflow for water distribution criticality analysis and proposes the Kolmogorov‐Smirnov distance statistic between solution sets as a measure of the significance of interdependency for decision making. Results from the case study suggest that as the magnitude of water infrastructure failure increases beyond a threshold, the interdependency between water distribution and transportation becomes more significant. The difference between identified critical components using only information from water distribution and using both water distribution and transportation is significantly different (with greater than 95% confidence) for the city of Tampa, when more than 40 components fail (are isolated). These results will assist utilities in asset management and strategy assessment, by helping prioritize component repair and better allocate resources for critical interdependent infrastructures. 

   more » « less
5. Stability and Multigroup Fairness in Ranking with Uncertain Predictions

   Devic, Siddartha; Korolova, Aleksandra; Kempe, David; Sharan, Vatsal (February 2024, Proceedings of the 41st International Conference on Machine Learning)

   Salakhutdinov, Ruslan; Kolter, Zico; Heller, Katherine; Weller, Adrian; Oliver, Nuria; Scarlett, Jonathan; Berkenkamp, Felix (Ed.)

   Rankings are ubiquitous across many applications, from search engines to hiring committees. In practice, many rankings are derived from the output of predictors. However, when predictors trained for classification tasks have intrinsic uncertainty, it is not obvious how this uncertainty should be represented in the derived rankings. Our work considers ranking functions: maps from individual predictions for a classification task to distributions over rankings. We focus on two aspects of ranking functions: stability to perturbations in predictions and fairness towards both individuals and subgroups. Not only is stability an important requirement for its own sake, but — as we show — it composes harmoniously with individual fairness in the sense of Dwork et al. (2012). While deterministic ranking functions cannot be stable aside from trivial scenarios, we show that the recently proposed uncertainty aware (UA) ranking functions of Singh et al. (2021) are stable. Our main result is that UA rankings also achieve group fairness through successful composition with multiaccurate or multicalibrated predictors. Our work demonstrates that UA rankings naturally interpolate between group and individual level fairness guarantees, while simultaneously satisfying stability guarantees important whenever machine-learned predictions are used. 

   more » « less