This R&D effort will develop a Stochastic Air Traffic Flow Management Rerouting Problem implementation on Graphics Processing Units. The primary application of this implementation will be for solving the nationwide TFM problem, while managing uncertainties associated with large-scale climate disruptions such as such as volcanic ash, other natural disaster phenomena and convective weather avoidance. The GPU implementation will exploit the Bender's decomposition for speeding up the solution. Bender's decomposition has been used by researchers to solve sequencing &scheduling problems in the terminal & transition airspaces, under NASA's Airspace Super Density Operations research focus area. The parallel Bender's decomposition implementation developed under this R&D effort can be used to speed up solutions to the terminal area sequencing and scheduling problems.
The Stochastic TFM formulation implemented on high- performance Graphics Processing Units can serve as an operational tool at FAA's Air Traffic Control System Command Center (ATCSCC) for managing air traffic in the NAS. Apart from air traffic management, stochastic programming finds wide applicability in various industries such as transportation finance and manufacturing. Stochastic Programming has been used for, 1) Solving supply chain network design with the uncertainties of processing/transportation costs, demands, supplies and capacities, 2) Cash management in automatic teller machines, 3) Restaurant revenue management, 4) Airline crew scheduling problem under uncertain schedule disruptions, and 5) Airline fleet composition problem. The generic stochastic programming solver developed under this R&D effort will be useful in solving large-scale stochastic programs in the above mentioned areas.