Stochastic Optimal Control for Ground-Based Metering Operations

Abstract

Trajectory-based operations is a NextGen program concept that uses trajectory prediction, scheduling tools, and improved methods for trajectory management to improve the capacity and efficiency of the National Airspace System. Metering operations are one element of trajectory-based operations and involve the sequencing and scheduling of aircraft to flow-constrained points or meter points. Aircraft trajectories are managed to the schedule to ensure aircraft-to-aircraft spacing across the meter point. Automation tools are being developed to support air traffic controller tasks in managing aircraft trajectories to meet scheduled times of arrival at the meter points. Uncertainties in trajectory prediction, such as those related to wind-forecast and aircraft-modeling errors, pose a significant challenge in the design of automation for trajectory management. This paper proposes a framework for exploring and designing metering strategies in the presence of trajectory uncertainties, in which tradeoffs between the costs of interventions and the costs of precisely meeting a scheduled time of arrival are quantified. A ground-based metering operation is formulated as a discrete-time, finite-horizon Markov decision process from which different control strategies for meeting a scheduled time of arrival can be explored. Simulation results demonstrate the opportunities for exploring control laws in a quantitative manner and the feasibility of designing optimal control laws given the trajectory uncertainties.