System of system design-for-resilience heuristics derived from forestry case study variants
Bryan C. Watson, Zack B Morris, Marc J. Weissburg, Bert Bras
Abstract
System of Systems connect complex technological systems (e.g. power plants, transportation, and infrastructure) to provide additional services. Failures in one constituent system, however, can often cascade to other systems within the network. Decision makers need repeatable, generalizable System of System design-for-resilience approaches. Current design-for-resilience approaches, however, are often not generalizable, do not fully leverage network structure as a design parameter, and require designers to identify possible faults before they occur. These gaps motivate our research question: How can biologically inspired design be used to increase System of Systems resilience to unexpected link failure? We hypothesize that examining a set of 20 possible design variants for the same case study will reveal correlations between Ecological Network Analysis metrics (a subset of graph theory) and increased resilience. These correlations are then used to identify 13 design-for-resilience heuristics. Heuristics are simplified design guidelines that improve, but do not necessarily optimize, decision making during the design process. This work provides two contributions. First, the heuristics provide an approach to both design for increased resilience and evaluate potential System of Systems designs. A discussion of when the heuristics are applicable is presented (i.e. initial graph theory metric value limitations). Secondly, validation tests provide evidence that the heuristics can be used to increase resilience for networks other than the case study. This article provides a crucial step by utilizing Ecological Network Analysis to identify and test a set of design-for-resilience heuristics, an important step to developing a repeatable, network-based, fault agnostic approach to design Systems of Systems with increased resilience.