Towards a Fast-time Simulation Analysis of Benefits of the Spot and Runway Departure Advisor

NASA Ames Research Center is developing a concept for managing flight operations on airport surfaces. The goal of the concept, named the Spot and Runway Departure Advisor (SARDA), is to reduce delays, emissions, noise, and fuel consumption. In 2010, human-inthe-loop simulations of the SARDA concept were conducted. Results showed that for the 2008 heavy traffic scenario SARDA reduced departure delays and fuel consumption, while imposing little impact on perceived controller workload. The 2008 normal traffic scenario did not show measurable benefits. The human-in-the-loop simulations analyzed only two traffic scenarios and were costly. To efficiently expand the research of SARDA so that it included analysis of more traffic scenarios, adaptation to more airports, and investigation of changes to the concept, a fast-time simulation capability was needed. This paper documents the first attempt at fast-time simulation analysis of the SARDA concept. A fast-time simulation of traffic at Dallas/Fort Worth International Airport being autonomously managed by a delay-optimal runway scheduler was conducted. Although the simulation was meant to analyze the benefits of the SARDA concept, there were differences between how traffic was managed in the fast-time simulation and the human-in-the-loop simulations. The differences were due to difficulties adapting the delay-optimal scheduler from the human-inthe-loop simulations to the fast-time simulation and lack of capabilities of the airport operations model. The main differences were that the human-in-the-loop simulations included runway crossings in the optimal departure schedules, while the fast-time simulation did not, and the human-in-the-loop simulations controlled flights at the spot, while the fasttime simulation controlled flights at the gate. Results of the delay-optimal fast-time simulation showed benefits for the 2008 normal traffic schedule. It produced 8% less average taxi time and 40% fewer stops than a simulation controlled by a first-come-firstserved scheduler. However, the delay-optimal simulation had the same average delays as the first-come-first-served simulation. In terms of equity of flight delays, the worst delayed flight in the delay-optimal simulation had one and a half minutes more delay than the worst delayed flight in the first-come-first-served simulation.