Detection and Quantification of Inefficiencies in the National Airspace System

Previous research has indicated that a significant weakness in the functioning of the National Airspace System (NAS) is a lack of adequate feedback to operators such as airline dispatchers and air traflic managers regarding the impacts of their decisions (Billings, 1997; Carlson, et al., 1996; Kerns, et al., 1999; Smith, et al., 1997; Wickens, et al., 1997). In particular, although dispatchers get anecdotal evidence about the air traffic problems that their flights encounter, such as cornerpost swaps and airborne holding, they do not get routine objective data about the frequency and impact of such air traffic mmatives on the llights that they plan. Similarly, air traffic managers make decisions about traffic flows during events like severe weather, but they do not receive any systematic feedback about how successful these plans were or about their costs to the airlines. This inadequacy is in part a hold-over from a system design that decomposed overall system performance into a set of somewhat independent subtasks, so that each person (such as a traffic manager 01 a dispatcher) could do his or her job adequately without too much knowledge or feedback about the performance of other parts of the system. It is also in part due to the fact that different organizations have historically collected different types of data, without any integration of these data sets. As the system has become more integrated, and as the airlines have been given greater flexibility under FAA rmt~atives such as the National Route Program, this lack of data integration and lack of feedback has become a much more important problem. In response to this need, a software system has been developed that integrales FAA data about planned and actual roulings with airline data about planned and actual costs (fuel consumptions and various time metrics such as departure time and time airborne). This soflware system, which is currently being Beta-tested by the Air Traffic Control System Command Center and by 4 airlines, provides post-operations analyses to air traffic managers and dispatchers about routine inefficiencies due to Ilight amcndmcnts. (For example, during one mouth, flights from Chicago to Atlanta departing I 115 Zulu for a particular airline encountered airborne holding 43% of the time, resulting in the consumption of 27% more fuel while airborne and 34% more time.) In addition to such quantitative data, the system provides matching map displays that show planned vs. actual routings. Finally, the software contains data mining tools that can look at the entire NAS for some lime period and help answer questions like: I. What flights routinely experience airborne holding the most and what arc the costs associated with such holding? 2. How often are low altitude departure routes used? How much more often could they be used in situations where there are significant departure delays?