Decision‐Making Analytics Using Plural Resilience Parameters for Adaptive Management of Complex Systems

Abstract

It is critical for complex systems to effectively recover, adapt, and reorganize after system disruptions. Common approaches for evaluating system resilience typically study single measures of performance at one time, such as with a single resilience curve. However, multiple measures of performance are needed for complex systems that involve many components, functions, and noncommensurate valuations of performance. Hence, this article presents a framework for: (1) modeling resilience for complex systems with competing measures of performance, and (2) modeling decision making for investing in these systems using multiple stakeholder perspectives and multicriteria decision analysis. This resilience framework, which is described and demonstrated in this article via a real-world case study, will be of interest to managers of complex systems, such as supply chains and large-scale infrastructure networks.

Infrastructure resilience curves: Performance measures and summary metrics

“…Some analysts deliberately include stakeholders in the definition of measures and metrics. Thekdi and Santos examined supply chains and used expert elicitation to weigh five performance indicators for five stakeholder categories [81]. Rochas et al established measures for district heating performance using an analytic hierarchy process and a panel of experts [51].…”

“…Modeling may be primarily focused on the infrastructure system Modeling should include representation of system demand Modeling likely includes representation of system demand; may be primarily focused on noninfrastructure considerations components are functional or non-functional, and the measure describes "number of functional components" (e.g., power buses [75], [124], cranes at a seaport [103], or nodes in an electrical [62] or water system [62], [125]). This category also applies to components with partial functionality (e.g., bridge loading capacity [21], [126]) or weighted aggregates (e.g., total power rating of connected generators [23], or volume of transportation system capacity [81]). Measures in this category are distinguished as independent of system demand.…”

“…To quantify district heating performance, Rochas et al used an analytic hierarchy process and a panel of experts-while stressing that operational analysis should incorporate a larger group of experts [51]. For a supply chain model, Thekdi and Santos considered relative weighting across five stakeholders and five performance indicators; they found that the value of investment alternatives can be widely variable [81]. Across this literature survey, scenario-dependent weighting is acknowledged but unexplored.…”

“…Performance and time thresholds capture transitions in stakeholders' valuation of system performance. Two types of thresholds were found in the literature: critical performance thresholds, below which performance is unacceptable [57], [72], [80], [81], [83], [88], [90], [107], [177], [193], [195], [196], [212], [224]- [228], [232]- [235], and recovery time thresholds, within which the system should achieve specific recovery objectives [57], [67], [72], [80], [83], [88], [100], [107], [131], [196], [215], [226], [235].…”

“…Reiger defined brittleness as a modification of cumulative impact: the integral below a critical performance threshold [227], [228]. Thekdi and Santos used a critical performance threshold to override other metrics-falling below the threshold drives evaluation to zero [81]. Thresholds indicate a transition in stakeholder value-they may also indicate a discontinuity for the application of summary metrics.…”