This research describes an investigation into the potential impact of alternative management control policies on air traffic flow. An experiment was conducted that analyzed the effect of four different arrival management conditions under two different goal conditions on computer-simulated traffic management decision making and on simulated performance. In the condition representing the least level of central arrival management, there was no central arrival management decision maker at all. Two of the arrival management conditions resembled “free scheduling” operations: in both, a central arrival manager provided information to support independent departure decisions by air carriers, but the two conditions differed on the type of information provided. The fourth condition represented a centrally controlled arrival management operation, where the arrival manager directly limited the number of departures the air carriers could make.Each arrival management condition was simulated under two different goal conditions. A baseline condition was simulated, where all decision makers (air carriers and central) altruistically tried to maximize the same system arrival performance relative to a common utility function. This baseline condition was contrasted with an experimental condition, where each decision maker tried to maximize his own individual performance relative to his own utility functions.The results showed that when all decision makers were working to maximize the same system performance, there were no performance diffierences attributable to the arrival management conditions. However, when each decision maker was working to maximize his own individual performance, the best performance (both system and individual) occurred in the two “free scheduling” conditions, with worse performance under central management on one extreme and no management on the other. The results also showed that with simulated decision makers a “free-market” allocation of airspace resources can result in decisions that yield a dynamic equilibrium that is close to optimum. In addition, these simulated decision makers were able to evolve synchronized operations with no direct inter-user communication. Finally, the results also suggest that some sources of information may be important for such synchronized operations to evolve, but are not needed for maintaining that synchrony.
Read full abstract