NETWORK BASED MODELS FOR AIR-TRAFFIC-CONTROL

Several aspects of air-traffic control fit into the framework of network optimization models with nonlinear and nonseparable objectives. We develop two network based models for air-traffic control over the U.S. airspace. The first model deals with the dual objectives of monitoring congestion over high altitude jet routes while minimizing transportation costs. The second deals with the optimal flow of departing traffic through major airports to avoid delays at the destination. Of particular interest is the first model, due to the high increase of traffic volume over the last few years and the resulting congestion and increased number of accidents. A prototype model is built using data from a control sector in the Midwest. By integrating recent advances in numerical optimization, network flow theory and supercomputing we were able to solve the resulting problem very efficiently, thus demonstrating its feasibility. Computational experiments with real data on a CRAY X-MP/24 supercomputer are discussed.