Nested modeling for evaluating hazards associated with off-nominal scenarios in the National Airspace System

Understanding the safety implications of any changes to the operations of National Airspace presents a host of challenges due to the highly complex and coupled nature of this system. The system's current state might not be the most efficient, but its safety features are well established and grounded in years of relatively successful experience. Technological advances in both aircraft and the supporting infrastructure promise great improvements in efficiency, but their successful implementation necessitates appropriate procedural changes. As a result, a comprehensive assessment of the hazards associated with the introduction of new technologies must involve modeling of interactions among aircraft behavior, supporting infrastructure, and the operational procedures. The traditional approach to risk assessment involves a top-level fault-tree-based representation of the hazards that relies on binary logic and thus focuses on the sole fact of occurrence of relevant events regardless of their relative timing. Event trees account for the consequences of the relative order of events, but they lack any embedded means of evaluating the associated likelihoods that need to be computed externally, usually by means of physics-based simulations. The latter, while increasingly realistic in capturing particular physical phenomena, are limited to describing only few relevant interactions in order to keep the overall complexity of representation tractable. This paper argues for the need of an intermediate layer of analysis that, on the one hand, has enough fidelity to capture time-dependent coupling among relevant entities of the system, and on the other hand, is compact enough to track a large number of those relevant entities at the same time. Stochastic Petri Nets are suggested as a candidate for this intermediate layer, and the application is presented of the suggested nested analysis to the conflict resolution between the merging flows of air traffic that uses an optimized profile descent approach. Stochastic Petri Nets are coupled with agent-based simulation, and the efficiency of the merging procedures and their sensitivity to wind conditions and the traffic patterns are analyzed.