Print this page

Dynamical Airspace Configuration

In the current National Airspace System (NAS), the airspace of the continental U.S. is divided into twenty Air Route Traffic Control Centers, each of which is then divided into several air traffic control (ATC) sectors. An airspace sector is the fundamental operational unit for the purpose of air traffic control. The air traffic within each sector is monitored by one or two air traffic controllers to ensure safe operations of aircraft.

As the Next Generation Air Transportation System (NextGen) required, the ATC system need to be capable of handling two or three time of current flight demand through this network. The research of Dynamical Airspace Configuration(DAC) focus on splitting the non-transitional en-route space into sectors with uniform workloads while maintaining or even increasing airspace safety.

A systematically algorithm based on graph theory was proposed corresponding to the DAC problem. The flow chart is shown in figure 1.

Figure 1. Flow chart of DAC algorithm

The en-route airspace is firstly modeled as a weighted graph, using flight plans and flight tracks which accurately represent the air route structure and air traffic of the NAS. In figure 3, the Atlanta center (ZTL) is shown as an illustration of graph modeling.


Figure 3. The graph model of ZTL

(red triangles: main airports; green circles: waypoints; and green lines: current sector boundaries)

Then we developed a method accounting for workload calculation, incorporating monitoring workloads, coordination workload and conflict avoidance workload, which are commonly recognized as three of the most important factors to determine the air traffic control complexity (dynamic density). From Figure 4, we can read that about 70% of the total flights during the two-hour period are within 4nm of the graph model, about 92% of the total flights are within 8nm, and only less than 1% of the total flights are deviated from the graph model more than 16nm, which demonstrates that the final graph model in Figure 3 accurately represents the main air traffic structure in ZTL.

Figure 4. A histogram of flight track deviations from the graph model in ZTL

obtained from two-hour ETMS data between 4pm and 6pm EST on March 27, 2007.


In the next step, the spectral clustering algorithm is implemented, and then the graph model is partitioned into certain numbers of parts. As an illustration in figure 5, the Atlanta center is partitioned in to 12 sectors, and different group of vertices are denoted by different colors.


Figure 5. A partition of ZTL with 12 sectors


Finally, we proposed a combinatorial method involving in computational geometry to build the physical sectors based on the partitioned graph. Figure 6 shows a caparison between one of our results(Red) and current static sectorization(Green) . The sectorization is corresponding to the partition in Figure 5.

Figure 6. Current sectors (Green) and our DAC sectors (red)


For more simulation, we test our algorithm through generation dynamical sectors on other four air traffic centers, which significantly demonstrate the performance of our method.

Figure 7. New sectorizations of four Centers generated by the proposed DAC algorithm:

(top left) Cleveland Center; (top right) Dallas Center;

(bottom left) Kansas Center; (bottom right) Denver Center

(black lines represent boundaries of new sectorization and green lines represent boundaries of current sectorization)


We also have done a lot further analysis based on our results. The most recent one is the benefit analysis based on Cleveland center. In consideration of altitude, we treat the air space as two part vertically: low sectors (10,000-24,000 feet) and high sectors(24,000 feet above). We calculate the weight matrix using every half hours (and every one hour) ETMS data and then generate the sectors correspondingly. Comparisons between the number of sectors and the number of aircraft in ZOB on Feb. 8th, 2009 is shown in Figure 8 and Figure 9.


Figure 8. High sectors

Figure 9. Low sectors

Previous page: Throughput Analysis
Next page: Conformance Monitoring