Dynamical-Network Evaluation and Design Tools for Strategic-to-Tactical Air Traffic Flow Management

Abstract:

The objective of this research is to develop tools for comprehensive design and optimization of air traffic flow management capabilities at multiple spatial and temporal resolutions: at a national airspace-wide scale and one-day time horizon (strategic time- frame); and at a regional scale (of one or a few Centers) and a two-hour time horizon (tactical time-frame).

The following results were obtained in Year 4 of the project:

1) The UNT group continued to explore optimal management strategy design in the presence of weather uncertainties. Firstly, we developed an innovative uncertainty evaluation method M-PCM-OFFD that is effective and scalable to address large-scale system applications that typically involve a large number of uncertain parameters. The method breaks the curse of dimensionality for multi-dimensional uncertainty. Secondly, we extended the jump-linear approach toward tractable design of multiple management strategies under uncertainty. Optimization constructs were smartly integrated. Thirdly, we developed a dynamic data-driven approach to select representative uncertain weather events for contingency planning. The selection captures the spatiotemporal correlation of data drawn from spatiotemporally evolving dynamic processes. All methods were evaluated using real datasets and successful technology transfer has been achieved through the collaboration with the MITRE Corporation.

2) The WSU group continued the effort to develop weather-impact models for strategic decision-making. Specifically, a new operationally-structured model was developed and tested, which predicts runway configuration and capacity over a 24-hour horizon. Rather than being based on data-theoretic approaches, this model was structured to replicate the operational decisions made by managers at airports. Core network- controls research is also being pursued in support of these objectives. The model's performance using real weather-forecast data was tested for seven major airports, including Boston, Chicago, Dallas, Houston, and the New York area airports; the model performed well, and is tentatively slated for deployment at the Air Traffic Control Strategic Command Center (ATCSCC). Additionally, the WSU group pursued theoretical research on the underlying network control theory, focusing particularly on characterizing disruptive impact from actuation at a small subset of network nodes.

3) At Purdue, a fast decision support algorithm was developed to integrate terminal airspace operations under uncertainty. A multistage stochastic programming approach was chosen to formulate the problem and candidate solutions were obtained by solving sample average approximation problems with finite sample size. A multithreading technique was introduced to handle extensive computations. A proof- of-concept study was performed on a model of the northern-western flows of the Los Angeles terminal airspace considering a fleet mix of 14 aircraft. The application of the algorithm shows that flight time savings can be obtained when an hybrid separation method that combines altitude separation rules and temporal controls is used to separated traffic even in the presence of uncertainty. To characterize the results, an uncertainty study was conducted and shows that a compromise between time and computation time can be found with 100 scenarios per stage.

We envision that these results address critical needs in the strategic-to-tactical traffic management in the national airspace system (NAS). Moreover, the analytical tools developed broadly permit the tight conjoining of cyber- and physical- resources in designing decision-support capabilities for infrastructure networks.