NAS System Performance

Schedule and Airspace Usage Optimization Modeling

The U.S. National Airspace System (NAS) operates as a complex air traffic network composed of airport terminal nodes connected by multiple en route airway paths. The most severe disruptions to NAS network operations are caused by summertime thunderstorms, which decrease available en route and/or terminal airspace capacity. This can result in large delays, diverted airborne flights, and flight cancellations.

gridlockAirport gridlock. (Photo: Bruce Leibowitz)

Methods for the quantitative assessment of NAS performance during periods of significant convective weather impact are needed to support the development and evaluation of new Weather–ATM decision support capabilities. One of the primary areas for assessment is resource utilization, defined by the FAA/airline Customer Perspective Metrics Working Group (CMWG) as “the safe and efficient use of available airport or airspace capacity.” To date, the CMWG has not developed a quantitative metric for measuring resource utilization.

Measurement of capacity utilization during convective weather is difficult because storms cause capacity reductions in both en route and terminal airspace regions. In particular, en route capacity loss results in network congestion that cannot be readily characterized by scalar metrics such as the ratio of overall demand to a single capacity number.

Bertsimas Model for Weather–ATM Capacity Utilization (WACU)

Lincoln Laboratory has been working with Professor Dimitris Bertsimas from the MIT Sloan School to develop the Weather–ATM Capacity Utilization (WACU) model, a NAS scheduling and airspace usage model that seeks to minimize the total flight delay in the model domain while taking into account both operational logistics and convective weather impacts. WACU is an extension of an integer programming (IP) model developed by Professor Bertsimas and Sarah Stock-Patterson in 1998. The model solution yields the minimal cost (in terms of airborne and ground delays) and the flight plan for each flight – takeoff and landing times, and arrival times at each sector along its path. Convective weather impacts input to WACU are generated by weather-impacted sector capacity models developed at Lincoln Laboratory.

The results of the actual vs. modeled comparisons of capacity usage can support a number of Weather–ATM related NAS investment and performance assessment issues:

  • Quantitative estimates for “resource utilization” metric
  • Business-case development for new Weather–ATM decision support capabilities
  • Next-day FAA/airline review and post-event assessment of implemented tactical/strategic Weather–ATM decisions
  • Quantitative studies of the network usage changes that would occur with proposed NAS capacity enhancements (e.g., airway modifications, new runways, etc.)
  • Improved measurement of ATM performance and greater airline/public awareness of unavoidable delay
  • Resource utilization ramifications of fleet mix changes or changes in air traffic demand
  • Improved traffic manager training that includes studies of actual “missed opportunity” scenarios identified in the comparisons

Estimation of Unavoidable Delay

One application of WACU that is currently being investigated is the estimation of “unavoidable delay” due to convective weather. Unavoidable delay is the delay that would be incurred even if decision makers had perfect knowledge of the weather and made optimal operational decisions. To calculate unavoidable delay, WACU uses sector capacity impacts based on actual (not forecast) weather. Figure 1 presents the results from three different case days, run over the complete CIWS domain. Since WACU outputs specific flight plans, it can be used to make very fine-grained comparisons between actual and optimal ATM outcomes. Figures 2 and 3 illustrate one example. In Figure 2, WACU outputs for departure streams from New York airports are compared to the actual departure traffic. Two departure flows – the North gates and West gates – were closed, but WACU shows them operating, albeit at reduced volumes as a result of convective weather in the region. Were those flows actually feasible? Figure 3 shows the same traffic flows 30 minutes later, operating with similar volumes through similar weather, suggesting that the WACU departure flows were, in fact, feasible. Figure 4 illustrates a second example, comparing WACU-generated and actual flight plans for three different flights.

Ongoing Research Efforts

Several areas of research are ongoing. Scheduled flight plan and re-routing models are being improved to increase their flexibility and operational fidelity. Efforts are also underway to make real-time operational use possible. These include improvements in computational performance, incorporation of weather and sector capacity forecast uncertainty and adaptation to support a “warm start” version that could be run at regular intervals in real time with updated schedule and weather information.

top of page