Sensitivity of NASA’s Spot and Runway Departure Advisor to Traffic Forecast Errors

Abstract

NASA has developed the Spot and Runway Departure Advisor (SARDA), which plans spot crossing times and runway sequences to more efficiently manage departures on an airport surface, and extensively studied the concept and algorithms in the context of Dallas/ Fort Worth International Airport. This paper reports on a fast-time simulation study of NASA’s SARDA concept at Charlotte-Douglas International Airport (CLT). Previous work validated the SARDA concept at CLT and other airports and provided evidence that SARDA could provide substantial fuel-savings benefits by reducing the occurrence of long departure queues observed during current-day operations. This paper investigates the sensitivity of SARDA’s performance to uncertainty, measuring performance in terms of the runway throughputs and runway departure queue lengths achieved by SARDA. SARDA’s departure reservoir management function uses forecasts for the departure demand, taxi times from gate to spot to runway, and runway capacity. Errors in these inputs can result in either reduced runway throughput or reduced departure queue reduction benefit. For example, under-estimating the taxi time to the runway could cause departures to reach the runway later than expected, which could result in zero demand at the runway and lost runway capacity. A series of fast-time simulation studies were performed to measure the robustness of the SARDA model. Results demonstrate the importance of the forecast quality but support the conclusion that SARDA would provide benefits in a realistic operating environment. Furthermore, some forecast errors result from variations between flights that will not be predictable. The study identified a potential modification to the SARDA algorithm that is shown to improve robustness to uncertainty. The modification effectively dynamically adjusts the target queue length in response to the current uncertainty level.