ANALYSIS OF S-TURN APPROACHES AT JOHN F. KENNEDY AIRPORT

Transcription

1 ANALYSIS OF S-TURN APPROACHES AT JOHN F. KENNEDY AIRPORT Sebastian D. Timar and Katy Griffin, Saab Sensis Corporation, Campbell, CA Sherry Borener, Federal Aviation Administration, Washington D.C. C. J. Knickerbocker, Saab Sensis Corporation, E. Syracuse, NY Abstract This study analyzed tracking data for arrival flights to John F. Kennedy International Airport (JFK) during the period of August 16 to August 21, 29 to identify s-turn maneuvers and to characterize s-turn trajectories. The s-turn maneuver appears in the tracking data as a sequence of turns alternating in heading range. S-turns can arise as an alternative path stretching technique to vectoring and holding. This study developed analysis algorithms and associated metrics to automatically assess the occurrence conditions and operational impact of s-turn trajectories, and determined they (1) exhibit terminal airspace transit times several minutes greater than non s-turn trajectories, (2) occur in all runwaymetering fix combinations, (3) occur in merging and in-trail flight conditions, (4) occur at numerous distances from the airport, (5) do not exhibit significantly different inter-flight time spacing at the runway, and (6) the lateral spatial extent of s-turn maneuvers can be quite large. Analysis remains to assess the separation of s-turn trajectories from others throughout the maneuver, and to identify and characterize other such flight maneuvers and behavior in terminal airspace. Introduction This work was conducted as part of a project funded by the Federal Aviation Administration (FAA) Aviation Safety (AVS) organization to develop flight tracking data analysis tools to automatically detect and measure anomalous flight behavior in the terminal airspace as a basis for proactive safety assessments. The first step was to characterize the terminal airspace operations and flight trajectory behavior to establish a baseline for detecting anomalies. The s-turns appeared in the flight tracking data as a sequence of turns successively alternating in heading range. S-turns can arise as an alternative path /12/$ IEEE 3C1-1 stretching technique to vectoring and holding. S-turns are typically used in congested airspaces, such as the tight arrival corridors around JFK, where the controllers have very little space to employ a path stretch maneuver to delay an aircraft [1]. The s-turn maneuver is issued by the controller to the pilot as a sequence of vectors. The controller issues a heading change to one path, then issues a heading change to a different path before the aircraft levels out, thus creating the s-turn. This study developed flight tracking data analysis algorithms and associated metrics to automatically (1) identify s-turn maneuvers in arrival flight trajectories, (2) analyze the s-turn maneuver characteristics, and (3) analyze the aggregate behavior of the trajectories exhibiting s-turns. The study applied the algorithms to analyze flight tracking data obtained from the ASDE-X surveillance system. The flight tracking data was fused with ASR9 radar data having a 4.5-second sampling rate, with aircraft position then translated to Cartesian coordinates centered at the JFK tower. The paper is organized as follows. The background section discusses previous work leading to this particular study, and previous work related to this study. The detection section discusses the approach for identifying trajectory s-turns. The characterization section discusses metrics for differentiating s-turn maneuvers. The operations section discusses s-turn maneuver occurrence conditions and operational impacts. The conclusions section summarizes the work and provides forwardlooking statements. Background Related work includes flight trajectory data analysis for the characterization and modeling of terminal airspace and airport surface traffic, and algorithms for short-term prediction of separation losses in the terminal airspace.

2 This work builds upon a legacy of Saab Sensis Corporation work in trajectory data analysis for air traffic characterization, modeling and control. Examples include the following. Reference [2] analyzes aircraft position-time data to develop statistical models of approach flight lateral distance deviation from the ILS centerline. This supports estimating the likelihood of near-simultaneous approaches to parallel runways at Memphis International Airport (MEM) and Lambert-St. Louis Airport (STL). Reference [3] analyzes MLAT aircraft position-time data, in conjunction with data describing airport geometry, airspace features, and aircraft engine type, to develop computational models for real-time estimation of aircraft arrival times to the runway threshold from points at and inside of the corner post fixes at STL. This supports airport surface and gate traffic management and planning. Reference [4] analyzes aircraft position-time data to classify and quantify aircraft surface transit delays and identify their causalities at JFK. This supports a departure advisor decision support tool in planning gate pushback times and aircraft sequencing at the movement area entry points. Similar work by other organizations includes Reference [5], which analyzes flight trajectory data to model the speed profiles of arrivals on final approach to runways at Los Angeles International Airport (LAX). With respect to the larger project goal of realtime anomaly detection, relevant prior work exists in the domain of terminal airspace conflict detection. Previous studies such as [6] and [7] develop or utilize algorithms to infer flight intent from flight tracking data, apply the flight intent information to predicting the trajectories of aircraft in the terminal airspace, and detecting potential losses of separation. Alert lead times, false alerts, and missed alerts characterize the performances of these systems. Detection This study developed arrival flight tracking data analysis algorithms to identify trajectory turn maneuvers, then to identify s-turn maneuvers among them. Trajectory turn maneuvers were detected according to minimum specified turn rate, duration, and range threshold criteria. This study used the following threshold values for these criteria: minimum turn rate of 1.3 degrees per minute, turn duration of 1 seconds, and a range of 5 degrees. Trajectory s-turns were detected among the turn maneuvers using time separation and heading range criteria. This study used the following threshold values for these criteria: trajectory turn pairs no more than 1 seconds apart and opposite in heading range. Figure 1 shows an example arrival trajectory exhibiting a sequence of 5 successive s-turn maneuvers. This particular arrival flight was estimated to land to JFK on 16 Aug, 29. Figure 1. Approach Trajectory Exhibiting 5 Successive S-Turn Maneuvers Turn start points are indicated by green diamonds, end points by red crosses. S-turns ranged from very mild wiggles to large-scale maneuvers comprising significant heading changes. In the specified s-turn detection method, the second turn of one s-turn may have been the first turn of the next s- turn. For example, the trajectory shown in Figure 1 exhibits 5 distinct s-turns: The second turn of s-turn 1 is the first turn of s-turn 2, the second turn of s-turn 2 is the first turn of s-turn 3, and so on. This flight exhibits a sequence of successive s-turns, which correspond to a continuous heading change throughout a significant portion of the terminal airspace trajectory. 3C1-2

3 Characterization The study develops a number of metrics to characterize and differentiate among s-turn maneuvers. The metrics included the quantity of s- turns per trajectory, the severity of an s-turn maneuver, and the symmetry of an s-turn maneuver. The severity metric attempted to measure the magnitude of the s-turn maneuver to differentiate a trajectory wiggle from a large-scale maneuver spanning extensive lateral airspace. The asymmetry metric attempted to measure the shape of the turn maneuver in order to differentiate s-turn maneuvers comprising (1) a small turn followed by a larger turn, (2) the true s-turn where a large turn is followed by another large turn, or (3) a large turn followed by a smaller turn. Each metric is presented in detail below. ranges of the first and second turns of the s-turn maneuver. To demonstrate application of the severity metric and how it differentiates the arrival trajectories, Figure 3 and Figure 4 depict the lateral profiles (zoomed in to the s-turn portion) of two arrival flight trajectories exemplifying low and high severity scores, respectively. Quantity This study characterized arrival trajectories by the quantity of s-turns they exhibited. Figure 2 depicts the distribution of the quantity of s-turn maneuvers per trajectory among the 3,572 JFK arrival flight trajectories analyzed. Figure 3. Arrival Flight Trajectory with Low S- Turn Severity Score Figure 2. Distribution of Number of S-Turn Maneuvers per Arrival Trajectory The results indicate 61 percent of the trajectories analyzed exhibited one or more s-turn maneuvers, and 32 percent exhibited two or more s-turns. Severity This study proposed a severity metric as the geometric mean of the absolute values of the heading Figure 4. Arrival Flight Trajectory with High S- Turn Severity Score 3C1-3

4 This low severity s-turn maneuver reflects minor trajectory changes prior to final approach. This high severity s-turn maneuver reflects a right-hand turn (approximately 41 degrees) followed by a left-hand turn (approximately 9 degrees). profiles of two arrival flights exemplifying low and high asymmetry scores. Figure 6 depicts a trajectory with the low s-turn asymmetry score. The trajectory exhibits a small, symmetrical trajectory wiggle on final approach. To demonstrate application of the severity metric to characterize arrival trajectories, Figure 5 shows the distribution of severity scores among the s- turns in the arrival trajectories analyzed in this study. Also indicated in the figure are the mean and the 2- and 3-standard deviation ranges of the severity scores S-Turn Severity Scores µ 2σ 3σ Number of Arrival Trajectories S-Turn Severity, Deg Figure 5. Distribution of S-Turn Severity Scores The results indicate the s-turns tended toward lower severities. The mean of severity scores was 39. degrees and the standard deviation of the severity scores was 27.5 degrees. Figure 6. Arrival Flight Trajectory With A Low S- Turn Asymmetry Score Figure 7 depicts trajectory with the high s-turn asymmetry score. The trajectory exhibits an asymmetrical s-turn with one large right hand turn exceeding 36 degrees of turn range, followed by a small corrective turn in the opposite direction. Asymmetry This study proposed an asymmetry metric as the difference between the turn ranges of the first and second turns of the s-turn maneuver. Trajectories with a near zero asymmetry score execute a true s- turn; that is, two turns in succession which are nearly identical in their ranges, thereby forming an S shape. Trajectories with a large nonzero asymmetry score execute a large turn followed by a small corrective turn, or a small turn followed by a larger turn. To demonstrate application of the asymmetry metric and how it differentiates the arrival trajectories, Figure 6 and Figure 7 depict the lateral Figure 7. Arrival Flight Trajectory with a High S- Turn Asymmetry Score 3C1-4

5 To demonstrate application of the asymmetry metric to characterize arrival trajectories, Figure 8 shows the distribution of asymmetry scores among the s-turns in the arrival trajectories analyzed in this study. 6 4 Flight Index:: S-Turn Asymmetry Scores µ 2σ 3σ y, nmi 2 Number of Arrival Trajectories S-Turn Asymmetry, Deg Figure 8. Distribution of S-Turn Asymmetry Scores The results indicate the majority of s-turns had asymmetry scores close to zero, suggesting they are fairly symmetric; hence are true s-turns. Severe, Symmetric S-Turns This study identified severe symmetric s-turns as those with the combination of a high severity score (beyond 2-standard deviations severity) and a near zero asymmetry score (within 1-standard deviations asymmetry). These are approximate characteristics of a pure S shape, differentiating from those flights simply making one large turn preceded or succeeded by a small adjustment turn. Figure 9 depicts the lateral profile of an arrival flight exemplifying a severe symmetric s-turn Severity Score: Asymmetry Score: x, nmi Figure 9. Trajectory Exhibiting a High Severity Symmetric S-Turn The trajectory exhibits an s-turn with a pure S shape. The trajectory has a high severity score of degrees and a low asymmetry score of 13.6 degrees. Summary This study characterized the occurrence of s- turn, non s-turn and severe-symmetric s-turn trajectories among the arrival trajectories. Analyzing each of the six days, percentages of arrival trajectories exhibiting s-turns ranged from 36.6 to 42.2 percent, exhibiting non severe-symmetric s-turns ranged from 52.5 to 58.7 percent, and exhibiting severe-symmetric s-turns ranged from 2.6 to 5.5 percent. Analyzing all of the six days, percentages of s-turn, non s-turn and severe-symmetric s-turn trajectories were, respectively, 4.2, 55.4 and 4.4 percent. Thus, the majority of flights were exhibiting s-turn maneuvers, with severe-symmetric s-turns comprising a small portion. Operations This section discusses assessment of the characteristics and impacts of s-turn approach trajectories on the terminal airspace surrounding the airport. Areas assessed in this study included (1) terminal airspace transit times, (2) occurrences by fix-runway combination and traffic level, (3) inter- 3C1-5

6 arrival times at the runway threshold, (4) maneuver airspace utilization, (5) maneuver locations of occurrence, (6) maneuver shape, and (7) separation conditions. The assessments were enabled by first associating each approach trajectory with a particular arrival metering fix and runway. Arrival metering fixes were estimated visually by comparing local waypoints with trajectories. The local airspace waypoint and runway threshold data were obtained from the FAA National Flight Data Center (NFDC) database. The arrival metering fixes were estimated as LENDY, CAMRN and CCC. These were not verified against standard operating procedures. Minimum lateral proximity was the criteria for associating each trajectory and particular trajectory point with an arrival metering fix and a runway threshold. Transit Time This study compared the terminal airspace transit times of non s-turn, s-turn, and severe symmetric s-turn arrival flight trajectories. The terminal airspace transit time was estimated as the difference between the time when the trajectory entered a 55 nautical mile (nmi) radius circle centered at the JFK airport, and the time when it crossed the threshold of its estimated landing runway. Figure 1 shows the transit time distributions and the mean transit time of each population. Figure 1. Terminal Airspace Transit Times for S- Turn and Non S-Turn Arrival Flight Trajectories For the non s-turn, s-turn, and severe symmetric s-turn arrival trajectory categories, Table 1 presents the minimum and the maximum mean transit times among each of the 6 days analyzed, and presents the mean transit time across all six days analyzed. Table 1. Mean Terminal Airspace Transit Times, Minutes Trajectory type Non s- turns Minimum among days Maximum among days All days S-turns Severesymmetric s-turns Across all 6-days analyzed, the study finds s- turn trajectories have a mean terminal airspace transit time 3.3 minutes greater than that of non s-turn trajectories, and severe symmetric s-turn trajectories exhibit a mean terminal airspace transit time 7. minutes greater than that of non s-turn trajectories. These are significant increases. The largest and smallest differences between the mean transit times of severe s-turn and non s-turn trajectories are 12.6 and 3.2 minutes, respectively. Considering the flights on a single day, namely August 16, 29, multiplying the difference between the s-turn and non-s-turn mean transit times (found to be 3. minutes) by the quantity of s-turn flights (found to be 362 flights) yields an estimated 186 minutes of excess flying time. This generates excess fuel burn and emissions. From a queuing-theory perspective, longer terminal transit times translate lower terminal airspace service rates, thereby reducing throughput [8]. Traffic Level This study investigated the correlation of s-turns with traffic level. This was assessed by comparing the number of s-turn maneuvers initiated with the number of arrival flight runway threshold crossings in 15-minute time periods. Figure 11 shows the average of the number of JFK arrivals and s-turn maneuvers in each time period throughout each 24-3C1-6

7 hour day among the 6-days of trajectory data evaluated. 1 9 Percentage of Arrival Flights Executing S-Turns By Fix-Runway Combination LENDY CAMRN CCC Quantity Average Arrival Landings and STurn Counts Mean Flight Count Mean Sturns Count Percentage of Flights, % Hour, GMT Figure 11. Number of JFK Runway Threshold Crossings and S-Turn Maneuvers in 15-Min Time Periods Throughout the Day The results indicate that s-turn maneuvers were executed consistently throughout each day, and closely followed arrival traffic demand. A correlation coefficient of.85 between the numbers of landings and s-turns confirms this. During peak traffic periods, the quantity of s-turns sharply increased to greater than the number of arrival flights, indicating that arrival flights were performing multiple s-turn maneuvers. This implies significant controller workload levels during these periods, as controllers communicate each s-turn maneuver to the pilot as a vector sequence. Fix-Runway Combination This study investigated the occurrence of s-turn maneuvers among particular arrival fix and runway combinations. Figure 12 shows the percentage of trajectories in each arrival fix runway group that were estimated to execute s-turns maneuvers. 13R 31L 4L 22R 13L 31R 4R 22L Runway Figure 12. Percentage of Arrival Flights Executing S-Turns for Each Fix-Runway Combination Figure 12 indicates s-turns were executed by flights landing to all active JFK runways (13L/R, 31L/R, 22L/R) via each observed arrival fix (LENDY, CAMRN, CCC). Out of the 3,572 trajectories analyzed, we found only 2 flights used runway 4L and no flights used runway 4R. Therefore no data are plotted for these runways. These results indicate that the s-turn maneuver was used across all arrival fix runway combinations and that the maneuver was most common among the aircraft on the LENDY-13R flight path. Flight Pair Condition This study investigated the use of s-turn maneuvers to facilitate merging aircraft from different arrival fixes to a common runway. This was accomplished by analyzing s-turn flights and their runway landing predecessors (estimated from runway threshold crossing time) to identify whether they were from different arrival fixes. Figure 13 shows the percentages of sequential runway landing flight pairs from the same arrival fix and from different arrival fixes for each runway, and the percentages of those flights in which an s-turn maneuver was executed. 3C1-7

8 The inter-flight time spacing between each runway predecessor and successor is the difference between the trajectories estimated runway threshold crossing times. Only flight pairs with time spacing less than 7 minutes were analyzed. This is 3 times the default time spacing of 14 seconds for a Small weight class category aircraft following a Heavy as specified in [9]. This yielded 3,324 flight pairs for analysis, more than half of which were to runway 22L, and a quarter of which were to runway 13L. Figure 14 depicts, by runway, the inter-flight time spacing means among the four categories of flight pairs. Figure 13. Percentages of Arrival Flights from the Same and Different Arrival Fixes (Left) and S- Turn Trajectories Among Them (Right) 7 6 Time Spacing Mean & Two Standard Deviations Range by Pair Types, Runway Non,Non Non,Sturn STurn,Non SturnSturn The results indicate that, among the arrival flight pairs landing to each runway, approximately 4 percent were from the same arrival fix, and approximately 6 percent were from different arrival fixes. This is somewhat in proportion with the 3 observed arrival fixes and the approximately even traffic distribution observed among them. The results also indicate that among the arrival flight pairs landing to the same runway from the same arrival fix, approximately 2 percent of the trailing flights exhibited at least one s-turn, and among the arrival flight pairs landing to the same runway from different arrival fixes, approximately 3 percent of the trailing flights exhibited at least one s- turn. Thus, s-turn maneuvers do not appear to be used more for merging applications than for pure in-trail separation applications. Inter-Flight Spacing This study investigated the impact of s-turn maneuvers on inter-flight time spacing at the runway threshold, an indicator of runway throughput. For the analysis, each landing flight pair was sorted into one of four categories according to the trajectory types of the runway predecessor and successor. The four categories of runway predecessor-successor trajectory types comprised (A) non-s-turn, non-sturn; (B) non-s-turn, s-turn; (C) s-turn, non-s-turn, and (D) s-turn, s-turn. Time, Min L 13R 22L 22R 31L 31R 4L Runway Figure 14. Per-Runway Inter-Flight Time Spacing Mean and Two Standard Deviations Range by Trajectory Pair The results indicate for the two most heavily trafficked runways 13L and 22L, the s-turn trajectory pair categories (Categories B, C, and D) had a mean spacing which was less than the non-s-turn trajectory pair category (Category A). This may be due to s- turns typically arising under high demand conditions, where there is typically less time between successive runway landings. Across the runways, however, the mean inter-flight time spacing does not exhibit a conclusive trend among the flight pair categories. Figure 15 depicts, by runway, the inter-flight time spacing standard deviations among the four categories of flight pairs. 3C1-8

9 Figure 16 shows the location of the s-turn maneuver convex polygons within the JFK terminal airspace for all the s-turns detected among the 3,572 arrival flights analyzed. Each s-turn maneuver is represented by a pink convex polygon. 6 S-Turns 6 nmi 5 nmi 4 4 nmi 3 nmi 2 LENDY 2 nmi 1 nmi CCC Figure 15. Per-Runway Inter-Flight Time Spacing Standard Deviation by Trajectory Pair y, nmi JFK The results indicate for the two most heavily trafficked runways 13L and 22L, the s-turn trajectory pair categories (Categories B, C, and D) tended to exhibit slightly less inter-flight spacing variability than the non-s-turn trajectory pair category (Category A). Again, this may be due to s-turns typically arising under high demand conditions. Across the runways, however, the inter-flight time spacing did not exhibit a conclusive trend among the flight pair categories. Airspace Utilization This study investigated the airspace utilization of s-turn maneuvers, to characterize where they typically occurred and the area they occupied. The approach was to estimate a convex hull bounding the lateral portion of the trajectory segment comprising the s-turn maneuver. Given a set of points in the plane, the convex hull is the smallest convex polygon that contains all of the points, and has a subset of the points as its vertices [1]. It may be identified via a gift wrapping algorithm or other means [11] CAMRN x, nmi Figure 16. Locations of All S-Turns for August 16 th 29 Trajectories The results illustrate that s-turn maneuvers were occurring throughout the terminal airspace. The distribution of s-turn maneuvers is limited to 3 corridors defined by arrival fixes LENDY, CCC, and CAMRN. The locations of the s-turn maneuver convex polygons were observed for different traffic days to observe daily variation. Figure 17 shows the location of only severe-symmetric s-turns for August 16th 29 flights. 3C1-9

10 6 Semi-Severe/Symmetric S-Turns for Single Day 6 nmi 12 S-Turns 4 5 nmi 4 nmi 1 y, nmi 2 LENDY 3 nmi 2 nmi 1 nmi JFK CCC Number of Arrival Trajectories CAMRN S-Turn Area, nmi x, nmi Figure 17. Location of Severe-Symmetric S-Turns for August 16 th 29 Trajectories The results indicate that among the trajectories via CAMRN, s-turns typically occurred just after the fix, and many were estimated to have been severesymmetric s-turns. Among trajectories via LENDY, s-turn maneuvers typically occurred prior to the arrival fix, and upon approach to the runway. Those s-turn maneuvers which occurred close to the airport were typically minor heading adjustments occurring before, during, or after the large starboard turn required to land to JFK arrival runways 13L/R. Since these s-turns were relatively small they did not fall within the severe-symmetric category. Among the trajectories via CCC, the majority of s-turn maneuvers occurred close to the airport upon approach to 22L/R. The distribution of the surface areas computed for the polygons estimated for the s-turn maneuvers is shown in Figure 18. Figure 18. Distribution of S-Turn Maneuver Estimated Surface Areas The results indicate a significant quantity of the s-turn maneuver estimated surface areas were less than 2.5 nmi. As a reference, the surface area encompassed by 6 nmi circle approximation of terminal airspace is 11,39.7 nmi squared. Separation This study performed a cursory investigation of aircraft separation as causality for initiating s-turn maneuvers. For each s-turn maneuver, the lateral and vertical separation of the trajectory with its runway predecessor at the s-turn maneuver start time was computed. The lateral separation data exhibit a tri-modal distribution, with modes arising in the ranges -2 nmi, 5-8 nmi, and 1-12 nmi. The first mode is dominated by s-turn trajectories landing to runway 22L, and the second and third modes by trajectories landing to runway 13L. The vertical separation data exhibited a single model distribution centered at approximately 5 ft, ranging between approximately -/+ 1, ft. Negative vertical separation values indicate the s-turn trajectory was at a lower altitude than its runway predecessor at the s- turn maneuver initiation time. Future analysis will compute the separation of the s-turn trajectory with spatially proximate trajectories at times prior to the s-turn maneuver 3C1-1

11 initiation time, and identify if separation minima are violated in that time. Conclusions In conclusion, this study developed and implemented algorithms to identify and characterize s-turn maneuvers in arrival flight trajectory data. The analysis was a first step in automatically detecting and measuring anomalous flight behavior in the terminal airspace as a basis for proactive safety assessments. The algorithms were applied to analyze a week s worth of arrival flight trajectory data for JFK airport during August 29. The study found s- turn maneuvers among more than half the trajectories analyzed. The maneuvers ranged from small flight corrections to large turns, identified as severesymmetric s-turns. The analysis determined s-turn trajectories exhibited a marked increase in terminal airspace transit time. The maneuvers correlated closely with traffic level, arose across all arrival fixrunway combinations, arose in merging and in-trail conditions, occurred throughout the airspace before and after the arrival fix, and did not appear to increase inter-arrival time spacing mean or standard deviation at the runway threshold. Follow-on s-turn studies include assessing the separation of s-turn flights from others in the terminal airspace throughout the s-turn maneuver, characterizing the altitude ranges where s-turns are accomplished, and characterizing the increase in controller workload due s-turns. This work is a preliminary but important step towards the goal of automatically detecting and measuring anomalous flight behavior in the terminal airspace as a basis for proactive safety assessments. Achieving the goal requires developing metrics to assess flight behavior, characterizing baseline behavior under different operational conditions and variables, and assessing historical or predicted flight trajectories against these baselines to identify anomalous behavior. The terminal airspace is a challenging domain for such a task given its numerous complexities, such as the lateral and vertical degrees of freedom of aircraft motion, use of vectoring in managing aircraft trajectories, runway configuration-dependent route structure, and multiairport metroplex interactions. RNAV/RNP procedures may facilitate aircraft trajectory assessment and anomaly detection by improving trajectory predictability and clarifying flight intent. References [1] Green, S., Vivona, R., Grace, M., Fang, T.-C., Field Evaluation of Descent Advisor Trajectory Prediction Accuracy for En-route Clearance Advisories, AIAA , AIAA Guidance, Navigation, and Control Conference and Exhibit, Boston, MA, Aug. 1-12, [2] Levy, B., Maximum Lateral Flight Deviations from the ILS Centerline, 24 th Digital Avionics Systems Conference, 3 October 25, Washington, D.C. [3] Levy, B., Bedada, S., A Real-Time ETA-To- Threshold Prediction Tool, 25 th Digital Avionics Systems Conference, 15 October 26, Portland, Oregon. [4] Levy, B., Knickerbocker, C., Stroiney, S., Ralbovsky, F., Analysis And Causality Of Airport Surface Delays, 28 th Digital Avionics Systems Conference, October 29, Dulles, Virginia. [5] Gong, C., Sadovsky, A., A Final Approach Trajectory Model for Current Operations, 1th AIAA Aviation Technology, Integration, and Operations (ATIO) Conference, September 21, Fort Worth, Texas. [6] Tang, H., Robinson, J., Denery, D., Tactical Conflict Detection In Terminal Airspace, 1th AIAA Aviation Technology, Integration, and Operations (ATIO) Conference, September 21, Fort Worth, Texas. [7] Kirk, D., Initial Functional Performance Assessment of A Terminal Airspace Conflict Probe Application, AIAA Guidance, Navigation, and Control Conference and Exhibit, August 23, Austin, Texas. [8] De Neufville, T., Odoni, A., Airport Systems Planning, Design, and Management, McGraw-Hill, New York, NY, 23. [9] Raytheon ACES Team, Airspace Concept Evaluation System (ACES) Follow-On Contract, Contract Number: NNA5BE1C, CDRL 19 3C1-11

12 System/Subsystem Design Description (SSDD) / Software Design Document (SDD), 15 November 25. [1] The Mathworks, Convex Hull, MATLAB Software User s Guide, 211. [11] Farouki, R., Pythagorean-Hodograph Curves, Algebra and Geometry Inseparable, Springer-Verlag, Berlin Heidelberg 28, pp Addresses Sebastian.Timar@saabsensis.com Katy.Griffin@saabsensis.com Collen.Knickerbocker@saabsensis.com Sherry.Borener@faa.gov 31st Digital Avionics Systems Conference October 16-2, 212 3C1-12