Analysis of Automated Aircraft Conflict Resolution and

Weather Avoidance

John F. Love* and William N. Chan.†

NASA Ames Research Center, Moffett Field, CA, 94035

Chu Han Lee‡

University of California Santa Cruz, Moffett Field, CA, 94035

This paper describes an integrated solution to traffic conflict detection and resolution
with weather avoidance in the Center-TRACON Automation System. An automatic conflict
resolution algorithm developed and tested to resolve aircraft-to-aircraft conflicts is modified
to reroute aircraft around convective weather constraints while avoiding traffic conflicts.
The approach extends algorithms which resolve aircraft-to-aircraft conflicts to also resolve
aircraft-to-forecast weather conflicts. Alternatively, a separate algorithm that solves aircraft
to weather conflicts would have to be integrated with the current method to solve both
aircraft and weather conflicts. This evaluation limited conflict resolutions to horizontal
maneuvers due to software limitations in the research software that are being addressed.
Simulation results show 96 percent of aircraft-to-aircraft conflicts without any weather
constraints were successfully resolved. In high traffic scenarios with moderate weather, 91
percent of conflicts were resolved, while in high traffic with bad weather 72 percent of
conflicts were resolved.

I. Introduction

I N order for Air Traffic Management Decision Support Tools based on real-time trajectory automation, such as the
Center-TRACON Automation System, to work in all weather conditions, convective weather information must be
integrated into aircraft traffic conflict detection and resolution automation. To address improving the usefulness of
the Center-TRACON Automation System in all weather conditions, this paper discusses the development of an
algorithm to automatically route aircraft around weather constraints while keeping them from conflicting with other
aircraft. Currently, the Center-TRACON Automation System does not account for convective weather in its tools
such as the Traffic Management Advisor1, Direct-To2 and in its conflict probe functionality. The FAA deployed the
Traffic Management Advisory across the country and the FAA’s Research, Engineering and Development Advisory
Committee recommends “…enhancing the Traffic Management Advisor to achieve a weather sensitive arrival
planning tool.”3
This study evaluates the extension of an algorithm to automatically resolve aircraft to aircraft conflicts, already
in the Center-TRACON Automation System, to also avoid weather conflicts. The approach described in this paper
was taken for the following reasons. Using a separate weather avoidance algorithm would require that it work with
the current aircraft conflict resolution algorithm in the Center-TRACON Automation System or an entirely new
algorithm that resolved both weather and aircraft conflicts be created. Another reason is to utilize the aircraft
resolution logic that was designed to create resolutions typical of the ones used by controllers as those resolutions
may be acceptable to solve weather and aircraft conflicts. This work will also evaluate the efficacy of this simple
extension to serve as a baseline of performance for other such algorithms implemented in the Center-TRACON
Automation System. The algorithm that was extended is based on auto-resolver logic that was developed by
Erzberger4 and used to evaluate weather avoidance logic without mitigating aircraft conflicts in the Airspace
Concept Evaluation System.5 Weather constraints for this study were defined by the Convective Weather
Avoidance Model6 which was integrated in the Center-TRACON Automation System. This weather model
*
Aerospace Engineer, Systems Modeling and Optimization Branch, Moffett Field MS 210-8, Member.
†
Chief, Aerospace High Density Operations Branch, Moffett Field MS 210-8, Senior Member.
‡
Senior Programmer/Analyst, University of California Santa Cruz/University Affiliated Research Center, Moffett
Field MS 210-8.

1
American Institute of Aeronautics and Astronautics
translates convective weather information to identify convective regions of airspace that pilots are likely to deviate
around.
This paper describes how the resolution algorithm was extended and the Convective Weather Avoidance Model
used for this study. A methodology section describes how the Center-TRACON Automation System was modified
to detect and resolve weather conflicts. Simulation results discuss the effectiveness of this algorithm to solve both
aircraft and weather conflicts. Finally, how the resolutions affect flight times for the simulation scenarios are
discussed. The paper ends with conclusions and suggestions for future work to improve tactical aircraft routing
considering weather constraints.

II. Simulation Methodology

The automatic aircraft-to-aircraft conflict detection and resolution module in Center-TRACON Automation
Software (CTAS) was modified to detect and resolve predicted aircraft encounters with weather. Regions of en
route airspace with convective weather that aircraft were likely to avoid were constraints. Weather constraints were
defined by the Convective Weather Avoidance Model (CWAM). Aircraft encounters with weather are treated the
same as aircraft conflicts with other aircraft.

A. Center-TRACON Automation System Software Baseline

CTAS is a suite of integrated decision support tools that uses 4-D trajectory modeling and analysis to provide
information and advisories to air traffic controllers. These advisories include conflict detection and resolution,
direct-to routings with one minute or more in savings, and airport arrival rate scheduling. The FAA version of the
Traffic Management Advisor (TMA),7 one element of CTAS, has been deployed by the FAA to all 20 en route
control centers but it lacks any ability to automatically detect and mitigate impacts to its computed schedule based
on convective weather constraints. The FAA version differs from the NASA research version used in this study but
their basic functionality remains the same. CTAS employs radar track and flight plan data from the FAA host
computer, wind data from the National Weather Service, and a database of aircraft type performance models. These
data allows the Trajectory Synthesizer (TS) module to integrate the equations-of-motion for each aircraft type to
create a 4-D trajectory. The 4-D trajectories are tested for both conflicts with other aircraft as well as intersection
with weather constraints that should be avoided. In this study, the aircraft automatically fly the 4-D trajectories that
were computed to be both conflict free of aircraft and weather.

B. The Convective Weather Avoidance Model

The Convective Weather Avoidance Model (CWAM) developed by MIT/Lincoln Laboratory is described in
DeLaura and Evans.8 CWAM is a statistically based model that translates convective weather information to
probabilities the pilot will deviate around weather. CWAM is based on the analysis of approximately 500 en route
flight trajectories through a single en route center from five different days in 2003 with significant convective
weather. DeLaura and Evans found that the difference between flight altitude and the radar echo top was the most
accurate predictor of aircraft deviation around convective weather. The vertical integrated liquid (VIL), which
measures precipitation intensity, played a secondary role. The Corridor Integrated Weather System (CIWS),9 also
developed by MIT/Lincoln Laboratory, is composed of echo tops and VIL data from weather radars and it provides
2-hour convective forecasts updated every 5 minutes with 5-minute forecast time-steps. The CIWS data are
processed by CWAM to create the probabilistic weather avoidance polygons that are a function of echo top height
and VIL intensity at each flight level from FL250 to FL450 ft. CWAM also creates forecast polygons based on the
CIWS forecast. Since CWAM has forecasts out to 2-hours, the predicted 4-D aircraft trajectories are time-
synchronized with the CWAM polygons when checking for conflicts. Time-synchronization is defined as
comparing the time step in the 4-D aircraft trajectory against the closet time-step of the CWAM polygon.
The avoidance logic is the same for the aircraft-to-aircraft resolutions with the additional constraint of having to
avoid CWAM polygons of 80 percent deviation probability. These probabilities were shown in a previous study to
be a reasonable surrogate for determining areas pilots would avoid flying through.8 Clearly, other deviation
probabilities can be used as Grabbe, et al.,10 found good results for their study of weather avoidance without
considering aircraft conflicts using 60 percent CWAM deviation probabilities.

C. Aircraft Conflict Resolution Algorithm

An automatic conflict resolution algorithm described by Erzberger11 that resolves aircraft-to-aircraft conflicts has
been integrated into CTAS. The algorithm incorporates procedures and heuristics that are typically used by air
traffic controllers to resolve conflicts in today’s operations. Aircraft conflicts are resolved one pair at a time using

2
American Institute of Aeronautics and Astronautics
altitude amendments, horizontal route amendments, and speed changes. The algorithm resolves all combinations of
conflicts between cruising, descending, and climbing aircraft. Conflict resolution maneuvers are designed to
minimize change in flying time delay relative to the original flight trajectory. The automatic conflict resolution
algorithm first checks a direct route to a downstream fix on an aircraft route of flight. These so called “Direct-To”
routes can often resolve traffic conflicts, save flying time for airspace users, and are operationally acceptable.2 If a
direct route does not resolve the conflict, the algorithm prioritizes the heuristic maneuvers for the problem being
solved and trial plans each maneuver to check for secondary aircraft conflicts until a resolution is found. The
separation criteria for avoiding other en route aircraft were 1000 ft vertical and 5 nautical miles horizontal.

D. Applying the Resolution Algorithm to Weather Conflicts

As a first approximation to an automated system of resolving both aircraft-to-aircraft and aircraft-to-weather
conflicts, the Advanced Airspace Concept resolution algorithm in CTAS was modified to treat trajectories crossing
into a CWAM polygon as head on aircraft conflicts. This method was chosen to explore if this simple extension
could be used to solve both aircraft and weather conflicts. If this algorithm cannot be extended, integrating a
separate weather avoidance algorithm would require that it work with the current aircraft to aircraft conflict resolver
or create a new aircraft and weather conflict resolution algorithm. This implementation abstracted a weather conflict
as a head-on conflict with the CWAM polygon because it does not favor either a right or left turn solution. The data
sent to the resolution algorithm is based on resolving a conflict at the point where the trajectory intersects the
CWAM polygon.
Only horizontal resolutions were used in this study. Altitude maneuvers were not used in this study for weather
avoidance because it was found that the version of CTAS used in this study would temporarily assign a lower
altitude sending aircraft under the CWAM polygons, which is not an acceptable solution. This limitation is
currently being addressed.
Figure 1 shows a flowchart for the iterative resolution process investigated for this study. There are two
detection modules, one for aircraft conflicts and another for weather. The CTAS en route module processes the
flight plan data and passes the position and velocity states to the TS to create the 4-D trajectory. The module then
passes the trajectory information to two parallel aircraft and weather detection modules. The aircraft detection
module looks for aircraft-to-aircraft conflicts predicted to occur within the next 20 minutes. The weather detection
module compares the aircraft’s next 20-minute trajectory with the forecast CWAM polygons.
The look-ahead time depends on the trajectory uncertainty, which is too large for conflict predictions beyond 20
minutes. CTAS’ software architecture is designed to function in a real-time radar tracking environment. CTAS
normally makes trajectory predictions based on the aircraft type and performance data. There can be differences
between radar track and calculated trajectories due to CTAS not having all the required information, such as the
weight of the aircraft. The difference between the actual and predicted aircraft locations results in uncertainty,
which grows with time. Restricting look-ahead times to 20 minutes is based on current practice to keep the
uncertainty within operational limits. This study uses recorded aircraft tracks but they are input to CTAS as if they
originated from an FAA radar host. In this simulation, the aircraft positions are determined by their recorded radar
tracks below 10,000 ft. Above this altitude, the flight states are determined by the predicted trajectory information
and there is no more uncertainty with the aircraft’s trajectory but other climbing and descending aircraft trajectories
may still depend on the recorded data.
If a conflict is detected, each detection module prioritizes the list of conflicts based on time to conflict and
chooses which to send on to the resolution algorithm module. Because of assumed trajectory uncertainty, the
detection module queues aircraft-to-aircraft conflicts on the list that are not predicted to occur within 8 minutes. The
aircraft-to-weather conflicts are queued until predicted to occur within 12 minutes. The weather conflicts were
given more time to reduce the turning angle required for resolution. Weather systems can be much larger than the 5
nautical mile separation criteria that the resolution algorithm was designed to avoid. Successful resolutions were
required to be clear of weather polygons but not by any specific distance. Assuming modern aircraft travel at 8
nautical miles per minute, beginning the search for resolutions at the 12-minute point means the algorithm begins
attempting to resolve the weather conflicts about 96 miles from the conflict.
The resolution algorithm applies the heuristics for conflict resolutions based on the particular problem. In the
case of an encounter with weather, the head on conflict geometry is passed to the algorithm. When a resolution was
generated, the algorithm checks for secondary conflicts with other aircraft or CWAM polygons. If no secondary
conflicts were found, the aircraft flight plan was amended, and the algorithm moves onto the next conflict received.
If no resolution was found in the current 12-second time cycle, the conflict was resent in the next time cycle. This
process continues until the time to conflict was below specified time limits. The limits were 3 minutes for aircraft-
to-aircraft conflicts and 1 minute for aircraft-to-weather conflicts. For conflicts involving weather, the algorithm

3
American Institute of Aeronautics and Astronautics
had the additional constraint that it could only change the path of the aircraft. This may have the side effect of
decreasing the success rate in congested airspace when reroutes cause secondary conflicts.

Figure 1. Flowchart of aircraft and weather conflict resolution process

4
American Institute of Aeronautics and Astronautics
E. Automatic Aircraft and Weather Resolution
Figures 2 through 5 show an example of an integrated aircraft/weather conflict resolution for a Northeast bound
aircraft at FL350. Shown in all the figures are the CTAS flight data block for the maneuvered aircraft, the flight
plan route, the high altitude sector boundaries in Indianapolis Center airspace, the CWAM weather polygons, and
other traffic. Figure 2 depicts the original flight plan trajectory. Other aircraft in the region appear as diamond and
triangle symbols, though not all are at the same altitude. The altitude dependent weather polygons are created by
CWAM representing an 80 percent probability of deviation. The intersection of the trajectory with the CWAM
polygon shown in Fig. 2 is more than 20 minutes away so it is not a detected weather conflict. Figure 3 shows the
aircraft at the same location and time but superimposed with the forecasted CWAM polygon 15 minutes in the

Conflict
detected

Figure 2. Example showing current aircraft position Figure 3. Example showing current aircraft position
and weather at time 0. at time 0 and forecast weather conflict at time t+15.

Figure 4. Updated aircraft position and weather at Figure 5. Aircraft’s position at time +5 and forecast
time +5 minutes. weather conflict at time +15 minutes.
future. The arrow points to the approximate location of the detected aircraft-to-weather conflict. By comparing
Figs. 2 and 3, the relative locations of the current and forecast weather polygons can be deduced by noting the
location of the sector boundary. The intersection of the aircraft trajectory and CWAM polygon in Fig. 2 is farther

5
American Institute of Aeronautics and Astronautics
away than in Fig. 3. The forecast conflict allows more time to detect and resolve weather conflicts rather than
waiting for an update to the weather at a future time.
Figures 4 and 5 illustrate the weather avoidance algorithm’s conflict free flight plan amendment. Figure 4 shows
the same aircraft at time plus 5 minutes. The aircraft is predicted to intercept the CWAM polygon in 10 minutes. At
this time, the resolution algorithm has found an acceptable solution free of secondary traffic and weather conflicts.
The aircraft’s flight plan is amended and the trajectory synthesizer module calculates a new 4-D trajectory, which is
displayed in Figure 4. Figure 5 is another view of the aircraft location at time plus 5 minutes with new flight plan
superimposed with the forecast weather the original trajectory would have intersected at time plus 15 minutes. This
is the weather conflict detected and sent to the resolution algorithm, which was resolved.

III. Simulation
For this study, 5.5 hours of high altitude traffic data from December 21, 2007 recorded in Indianapolis Center
(ZID) were split into two traffic levels (medium and high) and combined with three different weather scenarios.
Indianapolis Center was selected because the required weather data were available for this region. In the simulation,
aircraft would fly their original flight plan according to the radar track data and computed 4-D trajectories and
rerouted if a weather or aircraft conflict occurred. Data for all six runs were collected and compared. Total
aggregate airborne delay times in ZID center were also archived for evaluating the effect of rerouting.

A. Traffic Scenarios
The traffic and weather scenarios were chosen from different days so the traffic input to the simulation would
not be influenced by actual weather deviations. The date of the traffic data was chosen as Friday, December 21,
2007, from 0600 to 1130 Eastern Standard Time. This was a busy day at the beginning of the holiday travel season
with clear weather.
Figure 6 shows the recorded aircraft count for that date between the hours of 0600-1130 EST. This traffic file
provided both the medium and high traffic scenarios as the aircraft counts varied across this time period. The two
traffic level bands are annotated in Fig. 6. The medium traffic scenario was selected as the period from 0600 EST to
0830 EST, the high traffic scenario was selected as the period from 0845 EST to 1115 EST. The first 15 minutes of
the recorded data were removed after the simulation to eliminate cases where aircraft either initially start out inside a
CWAM polygon or the aircraft is already in conflict and there is insufficient time to find a resolution. For each
scenario, only 2 hours of data were analyzed. The medium traffic scenario started with around 100 aircraft and
increased to about 200 aircraft and the high traffic scenario started with over 250 and increased to about 340 aircraft.

Figure 6. Aircraft count versus time recorded in Indianapolis Center on December 21, 2007

6
American Institute of Aeronautics and Astronautics
B. Weather Scenario
Weather data were chosen based on the amount of convective weather coverage in ZID by inspection of
NEXRAD National Reflectivity Mosaics from the National Climatic Data Center. Based on these criteria, June 19,
2007 (bad), July 10th, 2007 (moderate), and July 24th, 2007 (none) were selected for the three weather scenarios.
The terms bad, moderate, and none refer to the overall area coverage.
Figures 7 and 8 show the radar reflectivity data
from the bad and moderate scenario, respectively,
within ZID airspace.12 The white square shown in the
figure is the approximate boundary of ZID airspace.
According to FAA Operational Network database, the
delays on these days were predominately due to
weather and not other factors such as equipment
problems, high volume of aircraft, runway closures, or
other issues.
No wind data were included in the simulation.
This study focused on resolutions so winds, though
not added, will affect the resolution results but does
not change the fact this algorithm can solve aircraft
and weather conflicts.
The CIWS radar data consists of an observation at
the current time and a two-hour forecast in 5-minute
Area Of Interest
increments. Although these data are updated every 5
minutes, a processing time lag was accounted for in Figure 7. June 19, 2007 2100Z radar reflectivity
the simulation so new weather data were limited to image on a “bad weather day”
every 15 minutes. Therefore, each two-hour
simulation included 8 CWAM weather updates, one
every 15 minutes.

IV. Results
The overall results for all six scenarios are
presented first. Then, a detailed look at the success
rates of aircraft-to-aircraft and aircraft-to-weather
conflicts along with the distribution of maneuvers is
presented. Next, the impacts of aircraft reroutes are
presented in terms of time savings and delays.
Finally, an example is used to illustrate the real-time
schedule impact of the weather rerouting.
Area Of Interest
A. Overall Results
Figure 9 shows the overall success rates of the six
traffic/weather scenarios investigated in this study. Figure 8. July 10, 2007 1700Z NEXRAD radar
There was no uncertainty due to forecast prediction reflectivity image on a “moderate weather day”
accuracy or aircraft trajectory uncertainty. The
aircraft automatically flew their conflict resolution advisories as proposed by the resolver. Success means that a
conflict resolution trajectory that resolves the detected traffic or weather conflict and does not create any other
traffic and weather conflict was found. The overall success rates, shown in green, include both aircraft-to-aircraft
and aircraft-to-weather conflicts. The success rates for aircraft-to-aircraft conflict resolutions without secondary
aircraft or weather conflicts remain above 96 percent for all traffic and weather scenarios. In the unresolved high
traffic, no weather, aircraft-to-aircraft conflicts, all eight were descending aircraft that had secondary conflicts. The
algorithm was able to resolve at least 91 percent of the conflicts under the moderate weather scenario for either
traffic levels. The bad weather scenario dropped the success rates down to about 72 percent.

7
American Institute of Aeronautics and Astronautics
 Figure 9. Simulation results for the traffic and weather scenarios

Table 1 shows the total number of conflicts detected during each scenario and corresponding success rate. For
the medium traffic and moderate weather scenario the total number of conflicts increased by 214 percent or more
than 3 times the no weather scenario. The bad weather scenario had more than 6 times as many aircraft and weather
conflicts as the no weather baseline. For the high traffic level, the numbers of moderate and bad weather conflicts
were 2 and 4 times the baseline, respectively.

Table 1. Individual and Combined Aircraft and Weather Conflict Resolution Success Rates
Traffic Weather Number Aircraft Aircraft Number Weather Weather Total Total
level level of conflicts success, of conflicts success, number success,
aircraft not percent weather not percent of percent
conflicts resolved conflicts resolved conflicts
resolved resolved

None 96 0 100 % - - - 96 100 %

Med. Moderate 62 0 100 % 220 17 93 % 299 94 %
Bad 69 0 100 % 375 162 70 % 606 73 %

None 219 8 96 % - - - 227 96 %

High Moderate 62 0 100 % 335 38 89 % 435 91 %
Bad 197 7 97 % 492 263 65 % 959 72%

Table 1 separates the aircraft-to-aircraft and aircraft-to-weather conflict resolution success rates. Notice that the
number of aircraft-to-aircraft conflicts decreases with the addition of moderate weather and increases for the bad
weather. Inspection of the resolutions revealed the weather rerouting eliminates some aircraft-to-aircraft conflicts.

8
American Institute of Aeronautics and Astronautics
The overall success rate decreases with increasing weather level because the algorithm can only change the flight
path of the aircraft. In an aircraft-to-aircraft conflict, the resolution algorithm can alter the flight path of either
aircraft. In congested airspace, the ability to modify either flight path may be advantageous when rerouting one of
the aircraft creates a secondary conflict and changing the other does not. The trial planner rejects resolutions that
result in conflict with either another aircraft or CWAM polygon. There are conflicts the resolution algorithm could
not solve. Other issues leading to the lower success rate as weather increases include the fact that the full set of
heuristic maneuvers available for resolving conflicts is not implemented for resolving weather conflicts. These
maneuvers include speed and compound horizontal reroutes with temporary altitude changes. At the time of this
writing, some of these maneuvers were ready for weather avoidance testing but still not available for this study.
Figures 10 and 11 show original and trial
flight plans generated by the iterative resolution
process described in Fig. 1 for achieving a
resolution that is free of traffic and weather
conflicts. The original flight trajectory is
shown in Fig. 10. The vertex and endpoint of
the trajectory line represent the two
downstream waypoints. Multiple CWAM
polygons and other traffic are depicted in the
vicinity. The other aircraft show are not
necessarily all the aircraft that could cause a
secondary conflict. The resolution algorithm
will select maneuvers from a list based on the
conflict geometry. The list is based on Figure 10. Original flight plan is in conflict with two
heuristics that mimic current pilot-controller weather polygons and other aircraft (square symbols).
operational practices. The list is also dependent
on aircraft type and performance. For weather
encounters, CTAS creates a head on conflict geometry at the location where the aircraft trajectory intercepts the
weather polygon. The resolution algorithm generates and checks maneuvers to avoid the conflict until it is resolved,
or all the maneuvers are attempted, or the process runs out of time. CTAS is designed around a real-time
architecture and each process has a defined amount of time to complete its task. Figure 11 shows the trial flight plan
trajectories. The algorithm generates small angle deviations before larger angles. In general, one vertex maneuvers
are attempted first, followed by two vertex maneuvers, then temporary altitude changes (altitude changes did not
work in the present study). One vertex maneuvers have one auxiliary waypoint before rejoining the original flight
plan whereas two vertex maneuvers have two auxiliary waypoints. The two vertex maneuvers result in a path
parallel and offset to the original flight plan and bypass the first downstream waypoint and rejoin the flight plan at
the second waypoint. In this case, no maneuver generated and checked for additional conflicts was able to resolve
the conflict.
Depending on the size and location of the
weather polygons and the number of other
aircraft in the vicinity, there may not be a
maneuver that can resolve the weather conflict
using the heuristics normally employed to
resolve aircraft-to-aircraft conflicts. The
resolution algorithm was designed to avoid
other aircraft that are many times smaller in
size than the weather polygons investigated in
this study. Information about the size of the
weather polygons is not transferred to the
algorithm. Additionally, the algorithm has
limitations in the size of the rerouting, mainly
that the total reroute should be less than 5
Figure 11. Original and proposed 4-D trajectories that
minutes of additional flying time and the
did not resolve a conflict with the two weather polygons
maximum distance between auxiliary
and other aircraft.
waypoints is 80 nautical miles. A parametric
study of these a priori defined options might

9
American Institute of Aeronautics and Astronautics
find better combinations that resolve more conflicts but care would have to be taken to ensure the resulting
maneuvers continue to be representative of current practice and their operational acceptance would remain high.

B. Distribution of Maneuvers
It was found that use of the automatic resolution logic for resolving conflicts between traffic and weather
sometimes resulted in aircraft being rerouted under the weather constraint. This is clearly an unacceptable weather
avoidance maneuver and future implementations must ensure that weather avoidance logic does not attempt to route
aircraft under weather cells. Out of a total of 2774 weather avoidance maneuvers in this analysis, 149 “successful”
resolutions were found to have been routed under weather and were removed from the results. These resolutions
were removed from all results.
Figure 12 shows the distribution of successful horizontal reroute maneuvers generated by the resolution
algorithm for both traffic and weather conflicts. The lines between the symbols are not intended to show a trend but
are there only to aid the reader in seeing the general trends as the one-degree angle increment makes the plot
cluttered, otherwise. Table 1 shows there were a total of 606 conflicts in this scenario but the total number of
maneuvers attempted to resolve the conflict exceeded 84,000. Figure 12 represents 444 successful resolutions. This

Figure 12. Number of successful left turn, right turn, and direct to waypoint maneuvers that
resolved conflicts for the medium traffic, bad weather scenario.

high total was accounted for by realizing that for each time cycle where an aircraft or weather conflict was not
resolved, the conflicts stays in the queue until it is either resolved or the time to conflict is less than the values in
Fig. 1. At the next time cycle, another attempt is made to resolve the conflict with new information. It is noted that
the resolution algorithm does attempt to use turn maneuvers greater than 90 degrees, however, most of the turning
solutions are less than 50 degrees. The maneuvers direct to a downstream waypoint of any turning angle are tallied
in the center of the plot at the zero degrees point. These maneuvers represent a reduction in the length of the flight
versus the planned route. Whether the maneuvers represent a time savings is dependent on the wind. The difference
in left versus right direct maneuvers is solely a function of the original flight plan and the conflict geometry.

10
American Institute of Aeronautics and Astronautics
C. Impact on Flights
Table 2 shows the time savings resulting from resolution maneuvers direct to a downstream waypoint. The

Table 2. Analysis of time savings for direct-to traffic and weather resolution
maneuvers only.
Direct to a Weather Time savings Time savings Time savings
Downstream Scenario No Weather Moderate Bad Weather
Waypoint Weather
Maximum 828 s 828 s 1315 s
Mean 62 s 75 s 109 s
Medium Median 28 s 37 s 47 s
Traffic Minimum 3s 3s 3s
Std Deviation ±29 s ±129 s ±238 s
Total Savings 45 m 59 m 121 m
Total Count 44 flights 47 flights 67 flights
Maximum 237 s 247 s 261 s
Mean 45 s 65 s 49 s
High Median 35 s 31 s 30 s
Traffic Minimum 0s 3s 0s
Std Deviation ±48 s ±67 s ±54 s
Total Savings 44 m 42 m 59 m
Total Count 57 flights 39 flights 72 flights

resolution algorithm always tests maneuver direct to a waypoint first in its hierarchy. Table 2 shows a trend of
maneuvers direct to downstream waypoint increasing time savings with increasing weather severity. This
simulation did not proactively search for direct maneuvers, it only searched when there was a conflict. All the
values shown in Table 2 saved time compared to the original route. Because wind data was not included in the
simulation, no direct routing to a downstream waypoint resulted in a delay.
Large time savings are attributed to aircraft filing a flight plan that allowed for short-cuts. However, these short
cuts did not take into account a wind field that may not have resulted in such large savings. This is supported in the
medium traffic scenarios by the 828 and 1315-second savings, 14 and 22 minutes, respectively. These flight plan
routes were generated while considering the weather and winds of Dec. 21, 2007. The original and amended
trajectories for both of these cases were investigated and it was noted that the original trajectory was significantly
longer. For the medium traffic scenario, the large 30 second time saving differences between the mean and median
values gives an indication of how much the mean was skewed toward the extreme values. There was also a
corresponding increase in the standard deviation for the medium traffic scenarios.
The high traffic scenario mean time savings were less than the medium scenario, suggesting that the weather
limited the unobstructed distance to downstream waypoint. The standard deviation is smaller than the medium
traffic and the total savings is less than the medium scenario.
The simulated scenarios investigated do not account for traffic flow management initiatives that may have been
in effect the day the flight plans were filed. Therefore, an increasing number of conflicts will lead to an increasing
possibility of finding a direct to maneuver with significant savings. The strategy of first looking for a direct
maneuver helps balance the total aggregate delay in these situations. In two out of the three high traffic scenarios,
the total number of aircraft benefiting from a direct maneuver increased. The maximum savings for the high traffic
cases were approximately equal and the standard deviation was much smaller for the moderate and bad weather
scenarios. The average savings ranged from 45 to 109 seconds, though some flights had little or no reduction in
flying time.
Table 3 shows the delays incurred as a result of all maneuvers other than direct-to resolutions. As expected,
aggregate delays increased with the severity of the weather. This delay increase ranged from 5 to 12 times when
compared to the no weather baseline although the mean delay per flight appears evenly distributed, remaining under
81 seconds. This shows that as more flights were affected, they all experienced a similar amount of delay. It was

11
American Institute of Aeronautics and Astronautics
also noted that some flights experienced a time savings even though the assigned maneuver was not directly to a
downstream waypoint. This occurs when a flight is given an auxiliary waypoint before rejoining the flight plan

Table 3. Analysis of delay incurred by rerouting of aircraft

Weather No Weather Moderate Bad Weather
Scenario  Weather
Minimum 11 s -50 s -114 s
Mean 48 s 70 s 81 s
Medium Median 39 s 40 s 56 s
Traffic Maximum 159 s 281 s 326 s
Std Deviation ±29 s ±50 s ±56 s
Total Delay 41 m 277 m 506 m
Total Count 52 flights 235 flights 377 flights
Minimum -233 s -6 s -203 s
Mean 45 s 67 s 69 s
High Median 39 s 40 s 40 s
Traffic Maximum 249 s 268 s 318 s
Std Deviation ±48 s ±44 s ±52 s
Total Delay 122 m 398 m 714 m
Total Count 162 flights 358 flights 617 flights

route but the total distance is still shorter than the original flight plan. The smaller standard deviations, as compared
to Table 2, shows the data are not spread out over the range. The difference in maximum delay amongst the
rerouted flights is under five minutes.

D. Rerouting Effects on TMA Schedule

With the weather translation model integrated into CTAS, it is now possible to show the impact the rerouting has
on TMA schedules. Since TMA is a tool within CTAS, it uses the same trajectory predictor. Therefore, changes in
an aircraft trajectory can be reflected as a change in schedule in the TMA timeline. Figure 13 shows a typical
weather conflict avoidance maneuver that is conflict free of weather and aircraft up to 20 minutes from the current
position of the aircraft. Figure 12 shows the original flight path trajectory of the SKW2605 aircraft, depicted as a
green line. The curved white lines show the TMA meter fix arcs and the white dots are numerous arrival fixes of the
destination airport. All other traffic has been filtered out. The aircraft is flying north to south and originates in the
Chicago ARTCC (ZAU) en route to a meter fix. High altitude airspace of ZID is shown in the center of the figure
with brown outlines. The flight data block of the aircraft is shown with altitude, ground speed, call sign, and aircraft
type. Altitude is in hundreds of feet and ground speed is in knots. The smaller polygons in Fig. 13 are CWAM
polygons with an 80 percent probability of deviation.
Figure 14 shows the successful conflict resolution trajectory path to avoid the CWAM polygon. The new
trajectory is clear of the weather constraints and other aircraft in the center airspace and the original meter fix still in
the flight plan.

12
American Institute of Aeronautics and Astronautics
Figure 13. Aircraft conflict with 80 percent CWAM contour
automatically detected.

Figure 14. Aircraft automatically rerouted to avoid the 80 percent

CWAM contour.

13
American Institute of Aeronautics and Astronautics
 Figure 15. TMA timeline before rerouting Figure 16. TMA timeline after rerouting
Figures 15 and 16 show the effect of the aircraft reroute around the weather using the TMA module in CTAS.
The bottom of the figure represents the current time in minutes after the hour at the meter fix and the aircraft and
numbers move toward the bottom as if on a ticker tape. When an aircraft is rerouted by a flight plan amendment, all
modules in CTAS are able to take advantage of the updated information. The SKW2605 flight shown in Fig 15 is
scheduled to arrive at the meter fix at 2 minutes after the hour. Figure 16 shows flight after rerouting around
weather. The aircraft is scheduled to arrive at the meter fix at 5 minutes after the hour. The delay for this rerouting
is about 3 minutes.

V. Conclusions
An automatic conflict resolution algorithm in the Center-TRACON Automation System for en route airspace
was extended to resolve traffic and convective weather conflicts in an integrated detection and resolution
methodology. This study evaluated the extension of an algorithm to automatically resolve aircraft to aircraft
conflicts already in the Center-TRACON Automation System to also avoid weather conflicts13. The approach
described in this paper was taken for the following reason. Using a separate weather avoidance algorithm would
require that it work with the current aircraft conflict resolution algorithm in the Center-TRACON Automation
System or an entirely new algorithm that resolved both weather and aircraft conflicts would need to be created.
Aircraft trajectories that were predicted to fly into weather constrained areas as defined by the Convective
Weather Avoidance Model were analyzed to find conflict free resolution trajectories in the same manner as other
aircraft-to-aircraft conflicts. The simulation combined two clear weather traffic scenarios with three weather
scenarios that increased from no weather to significant weather.
The success rate for resolving aircraft-to-aircraft conflicts without weather constraints is 96 percent or greater.
The four percent that were unresolved in the high traffic no weather scenario had secondary conflicts. Success rate
for high traffic with moderate weather was 91 percent while the high traffic with bad weather success rate dropped
to 72 percent. The success rate decreases with increasing convective weather because there may not be a maneuver
that can resolve the weather conflict using the heuristics normally employed to resolve aircraft-to-aircraft conflicts.
The success depends on the size and location of individual weather polygons, total area coverage, and the number of
other aircraft in the vicinity. Comparing the total number of conflicts to the no weather scenario, the bad weather
scenarios had six times and four times as many conflicts for the medium and high traffic simulations, respectively.
Individual and aggregate delays due to rerouting were analyzed. The aggregate delay almost tripled from 41
minutes to 122 minutes due to increasing traffic from the medium to the high scenarios without weather but the
mean delay was 48 seconds or less. The bad weather scenario increased the aggregate delay to 506 and 714 minutes
for the two traffic scenarios. The mean delay remained under 90 seconds in all scenarios for individual flights
affected and showed the rerouted flights experienced a similar delay as the number of flights affected increased.
Additionally, other flights experienced a mean savings ranging between 45-109 seconds because rerouting shortened
the flight path. Direct maneuvers always resulted in a savings because the simulation did not include wind.
Changes to the Traffic Management Advisor schedule resulting from automatically rerouting aircraft to avoid
both aircraft and weather was shown on the TMA timeline. The Traffic Management Advisor uses the same 4-D

14
American Institute of Aeronautics and Astronautics
trajectory predictor as other elements in CTAS. Therefore, software changes to allow visualization of aircraft flight
times on the TMA timeline were minimal as time steps are always computed and available.
This study was not intended to be a final solution in automatic resolutions of aircraft and weather conflicts but
the success rate shows the method has merit. Experience has been gained in integrating and utilizing weather
translation models in the Center-TRACON Automation System decision support tool. Implementing the full
complement of resolution maneuvers should improve the resolution algorithm’s performance.

VI. Future Work

Implementing temporary altitude, speed, and compound maneuvers will be the next step in improving the results
presented in this paper. Compound maneuvers consist of horizontal reroutes combined with temporary altitude
changes.
A future enhancement to the resolution algorithm could be to evaluate the size of gaps between weather
constraints to evaluate if those gaps are likely to be traversed by a flight. Closing off gaps that flights are unlikely to
traverse should improve system robustness by decreasing the number of rejected maneuvers.
Another option to be considered when a conflict cannot be resolved is constraint relaxation. This involved
increasing the allowable CWAM probability, i.e. from 80 percent to 90 percent, and minimizing the dwell time in
the CWAM polygon. If the conflict is successfully resolved with either option, the effect of the reroute on the
aircraft’s schedule should be analyzed.
If the improvements discussed above do not achieve the required success rates for an automated system,
integrating a separate weather avoidance that include a secondary aircraft conflict detection and resolution algorithm
may be required.

References
1
Hoang, T., and Swenson, H., "The Challenges of Field Testing the Traffic Management Advisor (TMA) in an Operational
Air Traffic Control Facility," NASA TM-112211, Aug. 1997.
2
McNally, B. D., Engelland, S., Bach, R., Chan, W., Brasil, C., Gong, C., Frey, J., Vincent, D., "Operational Evaluation of
the Direct-To Controller Tool," 4th USA/Europe Air Traffic Management R&D Seminar, Santa Fe, NM, USA, Dec. 3-7, 2001.
3
FAA REDAC, 2007, “Weather-Air Traffic Management Integration Final Report,” Weather –ATM Integration Working
Group (WAIWG) of the National Airspace System Operations Subcommittee, Federal Aviation Administration (FAA) Research,
Engineering and Development Advisory Committee (REDAC), Oct. 3, 2007.
4
Erzberger, H. (2006), “Automated Conflict Resolution for Air Traffic Control,” 25th International Congress of the
Aeronautical Sciences (ICAS), Hamburg, Germany, Sep. 3-7, 2006.
5
Windhorst, R. D. and Refai, M., “Convective Weather Avoidance with Uncertain Weather Forecasts”, IEEE/AIAA 28th
Digital Avionics Systems Conference, IEEE, 2009.
6
Chan, William, Mohamad Refai, Rich DeLaura: “Validation of a Model to Predict Pilot Penetrations of Convective
Weather”, 7th AIAA Conference on Aviation, Technology, Integration and Operations, Belfast, Ireland, Sep. 2007.
7
Hoang, T., and Swenson, H., "The Challenges of Field Testing the Traffic Management Advisor (TMA) in an Operational
Air Traffic Control Facility," NASA TM-112211, Aug. 1997.
8
DeLaura, R., and Evans, J., “An Exploratory Study of Modeling En Route Pilot Convective Storm Flight Deviation
Behavior,” Proceedings of the 12th Conference on Aviation, Range, and Aerospace Meteorology, Atlanta, 2006.
9
Klingle-Wilson, D., J. Evans: “Description of the Corridor Integrated Weather System (CIWS) Weather Products,” MIT
Lincoln Laboratory Project Report ATC-317, Aug. 1, 2005.
10
Grabbe, S, Sridhar, B., Mukherjee, A., “Integrated Traffic Flow Management Decision Making,” AIAA Meeting Papers on
Disc, Vol. 14 No. 9 (GNC/AFM/MST), AIAA, Reston, VA, 2009.
11
Erzberger, H., “Automated Conflict Resolution for Air Traffic Control,” 25th International Congress of the Aeronautical
Sciences (ICAS), Hamburg, Germany, Sep. 3-7, 2006.
12
Anon., “NEXRAD National Reflectivity Mosaics,” [Online database] URL http://www4.ncdc.noaa.gov/cgi-
win/wwcgi.dll?wwNexrad~Images2, National Climatic Data Center, Ashville, NC, 2009.
13
McNally, D., and Thipphavong, D., "Automated Separation Assurance in the Presence of Uncertainty," International
Council for the Aeronautical Sciences (ICAS) 2008 Congress, Anchorage, AK, Sep. 14-19, 2008.

15
American Institute of Aeronautics and Astronautics