1. Measurements of Wake Vortex Separation at High Arrival Rates: Proposal for a New WV Separation Philosophy
5th Eurocontrol/FAA ATM R&D Conference
23-27 June 2003, Budapest, Hungary
G. L. Donohue and R. C. Haynie
Dept. of Systems Engineering & Operations Research
George Mason University
Fairfax, VA
David Rutishauser
NASA Langley Research Center
Hampton, VA

2. (Departures / Hull Loss)
Research Motivation
Is there a safety-capacity trade-off?
What happens to Aircraft Separation Safety during periods of High Runway Utilization?
Should the Wake Vortex Separation Criteria and Separation Assurance System be Modernized?
More Safe
Hypothesized
Curve
Safety
(Departures / Hull Loss)
Less Safe
Capacity
(Departures / Year)
Low
High

3. Outline Current Static WV Separation Standards Data collection process
Results from data collection
Observations on Current Operations
Proposed Future Dynamic WV Separation Architecture
Summary and Proposed Work to Be Done

4. Current Static Wake Vortex Separation Standards
Small
4 Nm
6 Nm
Small
Large
5 Nm
4 Nm
5 Nm
Large
B757
Heavy
5 Nm
4 Nm
Small
Heavy
Heavy > 255,000 lbs
Large 41,000 lbs to 255,000 lbs
Small < 41,000 lbs

5. Typical WV Separation Times

6. Atlanta Airport: High Runway Utilization
2 Runways – Arrivals
2 Runways – Departures
50 Arrivals / Hr / RW – Max
72 Seconds between Arrivals
3.1% – Operations Delayed
(> 15 min)
Total Operations,
VMC, (per 15 min)
Airport Capacity Benchmark Report, FAA, 2001.

7. ATL Arrival - Departure IMC
120
ASPM - April Instrument Approaches
84,90
Calculated IMC Capacity
100
Reduced Rate (ATL) 80
Arrivals per Hour 60 40 20 20 40 60 80
100
120
Departures per Hour

8. Data Collection, Atlanta
Observation Point
Renaissance Hotel
One Hartsfield Centre Parkway
Atlanta, GA X
Runway 26
Runway 27
ATL Airport
Haynie, R.C Ph.D. Dissertation, George Mason University.

9. Data Collection Summary
Haynie, R.C Ph.D. Dissertation, George Mason University.

10. Data Collection Process
Airplane i
Threshold
Airplane i+1
Runway
. . .
. . .
. . .
Haynie, R.C Ph.D. Dissertation, George Mason University.

11. Data Manipulation 45 sec 108 sec – 77 sec = +31 sec Runway Occupancy
Time (RTI)
45 sec
Relative Inter-Arrival
Time
108 sec – 77 sec = +31 sec
Inter-Arrival
Time
Wake Vortex Separation Standard
Large following Large (3 Nm)
(3 Nm / (140 knots / 3600 sec/hr))

12. Runway Occupancy Time Runway Occupancy Time Normal(47.71, 8.33)
Observed data
Frequency
Normal fit
The best fit for the distribution of runway occupancy time of all types of aircraft is a Normal distribution. But we know the runway occupancy time is different for different type of aircraft, so we fit the distribution for each type of aircraft.
Runway Occupancy Time
Normal(47.71, 8.33)

13. LTI and ROT from the Observation and the Simulation
Simulation Result
ROT
ROT
Probability
To see whether the observation was reproduced correctly by the model, we put the two sets of data together and visualized them. It’s shown that the two distributions of LTI are almost the same except the simulation result is smoother. As for the runway occupancy time, the simulation result is more concentrated on the mean value which is around 48 seconds. The overall effect is satisfying.
LTI
LTI
Time (seconds)
Time (seconds)
LTI: Landing Time Interval; ROT: Runway Occupancy Time

14. Measured Separation in the WV Time Domain
Atlanta Runway 27
Collection Day #1, VMC
Total Observations: 103
Arrivals / Hr: 31
-100
-50 50
100
150
200
250
300
350
Relative
Inter-Arrival
Time
Observation #
(3.3 hours collection time)
Target
Haynie, R.C Ph.D. Dissertation, George Mason University.

15. Histogram of WV Time Domain Separation
Atlanta Runway 27
All Collection Days, VMC
Lost Safety
Lost Capacity
# Occurrences
Relative Inter-arrival Time (sec)
Haynie, R.C Ph.D. Dissertation, George Mason University.

16. ATL Histogram Composite: 3 Data Collection Periods
Perfect WVSS Adherence Value = 0 20 15
D1P1 36 Arr/Hr 10
D1P1 39 Arr/Hr
# Occurrences
D2/P2 39 Arr/Hr 5 20 40 60 80
-60
-40
-20
100
120
140
Relative Inter-Arrival Time
(sec)
Haynie, R.C Ph.D. Dissertation, George Mason University.
Bin size: 20 seconds

17. LGA WV Time Domain Data (VMC)
LaGuardia
Collection Day #2, VMC
Total Observations: 169
Arrivals / Hr:
-100
-50 50
100
150
200
250
300
Relative
Inter-Arrival
Time
Observation #
(5 hours collection time)
Target
Haynie, R.C Ph.D. Dissertation, George Mason University.

18. LGA WV Time Domain Data (IMC)
LaGuardia
Collection Day #3, IMC
Total Observations: 126
Arrivals / Hr: 34
-100
-75
-50
-25 25 50 75
100
125
150
175
200
Relative
Inter-Arrival
Time
Observation #
(3.7 hours collection time)
Target
Haynie, R.C Ph.D. Dissertation, George Mason University.

19. Comparison of Airports: Consistent Distributions
LGA in VMC -2 2 4 6 8 10 12 14 50
100
150
200
250
LGA in IMC
ATL in VMC
ATL in VMC
Aircraft
Per Runway
Per Hour
Inter-Arrival Time (sec)
Haynie, R.C Ph.D. Dissertation, George Mason University.

20. Baltimore IMC (18.7 Arrivals / Hr)
LGA & BWI Comparison
LaGuardia VMC / IMC
(4 periods, 27 – 34 Arrivals / Hr)
Target
Baltimore IMC (18.7 Arrivals / Hr) 10
VFR 33.8 Arr/Hr 8
IFR 34 Arr/Hr 6
VFR 30.9 Arr/Hr
Observations
VFR 27 Arr/Hr 4
IFR 18.7 Arr/Hr 2 20 60
-60
-20
100
140
180
220
260
Relative Inter-Arrival Time (sec)
Haynie, R.C Ph.D. Dissertation, George Mason University.

21. Safety / Capacity Relationship
ATL, DCA, LGA Historical Reports
# Reports
Filed
Percent Capacity Used
Haynie, R.C Ph.D. Dissertation, George Mason University.

22. Observations on Current Operations
Inter-arrival times indicate frequent loss of WV separation
Shape of inter-arrival distributions similar
For different airports
For IMC / VMC
Some evidence for decline in safety for higher arrival rates
Are Current Static Separation Criteria Correct for Capacity Constrained Airports?
Small data set (~ 1,500 points) – more needed
Data can be used as input to more sophisticated safety models (TOPAZ)

23. Is It Time to Change the WV Separation Philosophy?
Wake separation rules are static
Based on empirical measurements
represent a response to worst-case persistence of wake hazard
Over 30 years of wake research have produced the potential for a dramatic increase in knowledge about the persistence of wake hazard
New Data and technologies demonstrated in both the European and NASA Programs
Introduction of systems and procedures that utilize this improved knowledge of wake hazard durations Could allow for Increases in Capacity at a Specified Level of Safety!

24. Proposed Changes to Current Standards and Operations
Utilize a hybrid of ground-based and airborne systems
integrated to produce increased (and dynamic) knowledge and prediction of wake hazards
Develop and Deploy a System to provide accurate wake hazard durations for Arrival Rate Prediction
Controllers use hazard information to set Arrival Rate
Information also provided to pilots of appropriately equipped aircraft to enhance situational awareness
Aircraft Equipped to Maintain Time-Based Separation have Separation Responsibility
Separation responsibility remains with controller during instrument operations for Non-Equipped Aircraft

25. Proposed Changes to Current Responsibilities
Roles/responsibilities
System provides wake hazard durations/wake-safe spacing recommendations
Approach and Tower Controllers responsible for safe spacing/separation Arrival or Departure Rate based on system recommendations
Airport arrival/departure rate changes communicated to the appropriate traffic flow management entities (e.g. Air Traffic Control System Command Center (ATCSCC) and Traffic Management Units (TMUs)

26. Candidate System Architecture

27. Improved Terminal Weather Prediction Component
Current Airport weather system Should be augmented with wake and weather sensors and prediction algorithms
Wake algorithm provides probabilistic wake behavior output
Terminal Area micro scale weather prediction for wake hazard durations
Fusing algorithm combines sensor data and closes a feedback loop between wake and weather predictions and measurements

28. Candidate Modifications to Terminal Airspace Design
Define Appropriate Region of Protected Airspace
For runway configuration and operation targeted (single runway or multi-runway complex; approaches and departures)
Closed-Loop Prediction System
senses current conditions diverging from predictions and adjusts to more conservative spacing (e.g. 30 minute adjustments)

29. Candidate US Airports for a Dynamic Wake Vortex Separation System
ATL
BOS
DFW
EWR
LAX
LGA
ORD
SFO
JFK
MIA
MEM
CLT
Configuration
2 pair
Closely Spaced Parallel Runways (CSPRs)
CSPR &
Int.
CSPRs
CSPR & Int.
indep
1 pair
&
indep& int.
Operation to
Test
Single-rwy
Arr. &
Dep.
Arr. & Dep.
CSPR
Single-rwy & Int.
Single- rwy
% B757 &
Heavy 22 13 11 12 21 9 10 25 31
% hours below VMC for CY2000 35 34 55 39 49 28
Here is a candidate list of airports to test WakeVAS implementations. The traffic mix and IFR vs. VFR are shown as values for two solution space parameters of interest.
More data collection and research/modeling will be required to fill in all the parameter ranges and typical values from the last slide for these airports

30. Summary CONOPS Ref: Policy Change Requirements
Develop Consensus on Wake Hazard Definition and Target Level of Safety Desired
Amend Current wake separation rules to incorporate Dynamic, Weather-Dependent Spacing
System Development Requirements
Standards for aircraft weather data, and data links
Wake and weather sensor maturation/FSD
Closed-Loop, probabilistic wake predictor design/PSD
Interface design/PSD
CONOPS Ref:
NASA/TM , April 2003

31. Specific Work Areas
Quantify accuracy/performance of all subsystems (wake/weather sensors)
Development of probabilistic wake predictor
Quantify temporal and spatial variation of relevant weather parameters (impacts weather sensor placement and coverage)
Perform formal safety analysis; rare event quantification
Define and acquire consensus on wake hazard specification
Quantify system valid prediction durations
Quantify dynamic spacing impacts on NAS
Quantify pilot/controller workload issues
Display design
Define data link requirements
Obtain required resolution weather data

32. Backup Slides
Backup Slides

33. Current Operations Terminal Configuration >
Single Runway; Parallel Runways < 2500’ separation
Intersecting Runways
Departures
Behind B757 or Heavy – 2 min hold; 3 min if
intersection or opposite direction same runway, OR
Radar separation minima
Heavy behind heavy- 4mi
Large/Heavy behind B757 –4mi
Small behind B757 – 5mi
Large behind heavy – 5mi
Small behind heavy – 5mi 
For pairs not listed the separation is 3 miles
2 min behind B757 or heavy departure or landing if projected flight paths will cross; includes parallel runways more than 2500’ in separation if will fly through the airborne path of other aircraft
Arrivals
Radar separation minima (at threshold):
Small behind large – 4mi
Small behind heavy – 6mi
For pairs not listed the separation is 3 miles,
except 2.5 miles in cases when 50 second
runway occupancy time is documented
Non-Radar Minima: 2 min behind Heavy/B757
except for small follower, 3 min
2 min for aircraft arriving after a departing or arriving Heavy/B757 if arrival will fly through airborne path of other aircraft

34. 15 Min Arrival Rates Atlanta Airport Collection Day #1, VMC
15 min averages
11:15am
1:15pm
6:45pm
Arrivals
per
15 min
8:00am
10:30am
1:00pm
3:30pm
6:00pm
Individual
Flight Data
Average: 31 / hr

35. Arrival Rates Inter-Arrival Times During “Busy” 15-min Intervals
0.261
0.074
0.005
0.267
0.0345
0.153
0.148
0.0394
Inter-Arrival Times During “Busy” 15-min Intervals
(> 31 arrivals / hr)
Frequency
Inter-Arrival Time (sec)
0.0353
0.106
0.306
0.282
0.176
0.0588
0.0118
Inter-Arrival Times During “Light” 15-min Intervals
(< 31 arrivals / hr)
Frequency
Inter-Arrival Time (sec)

36. Modeling Implications
Typical Modeling Approach
Safety is a constraint (Maximum rate through node in network)
Capacity is metric of interest
Queueing Constraints
(Minimum Times Between Operations)
New Approach
Safety is a function of capacity / demand

37. Average Approach Speeds