Robust Air Traffic Flow Management: An Optimization-Based Approach

Abstract

Air Traffic Flow Management (TFM) personnel make flight delay and re-routing decisions to maintain the predicted airport and airspace-sector traffic demand at future time-periods at or below the respective resource’s predicted capacity. Because the traffic demand forecasts and capacity forecasts involve a high degree of uncertainty, TFM personnel often make conservative decisions that induce unnecessary delays in the National Airspace System (NAS). The FAA, NASA, and other organizations have been researching and developing optimization-based TFM decision support tools (DSTs) that are expected to reduce such unnecessary TFM delays. The Bertsimas Stock-Patterson (BSP) algorithm formulates the TFM problem as an Integer Program and provides a solution that minimizes NAS-wide TFM delays while keeping the traffic demand under capacity. The BSP algorithm computes delays to resolve all predicted capacity overloads based on deterministic traffic demand and capacity forecasts. In its current form, the algorithm is expected to be highly sensitive to uncertainties in demand and capacity forecasts. This paper presents a Monte Carlo simulation-based approach for quantifying the sensitivity of the BSP algorithm to uncertainties, and assesses candidate algorithm enhancements for making the BSP TFM solutions robust to uncertainties. Simulation results show that the BSP algorithm is more sensitive to airport capacity prediction errors than to departure time prediction errors. Results also show that robustness-enhancements to the BSP algorithm using stochastic traffic demand and capacity prediction data can significantly reduce the NAS delays while successfully resolving all congestion events.