recent nasa wake surfing flight research
TRANSCRIPT
Recent NASA Wake Surfing Flight Research
EXPERIMENTAL MEASUREMENTS OF FUEL SAVINGS DURING AIRCRAFT WAKE SURFING
C U R T H A N S O N , J O E PA H L E , J A M E S R E Y N O L D S , S T E P H A N I E A N D R A D E , N E L S O N B R O W N
N A S A A R M S T R O N G F L I G H T R E S E A R C H C E N T E R
EXPERIMENTAL MEASUREMENTS OF PASSENGER RIDE QUALITY DURING AIRCRAFT WAKE SURFING
C U R T H A N S O N , S T E P H A N I E A N D R A D E , J O E PA H L E
N A S A A R M S T R O N G F L I G H T R E S E A R C H C E N T E R
introduction
1
Wake Surfing Background
WAKE SURFING EXTRACTS ENERGY FROM THE
UPWASH OF ANOTHER AIRCRAFT’S WAKE VORTEX.
Wieselsberger, C., “Contribution to the Explanation of Angled Flight Patterns of Some Migratory Birds,” 1914.
REDUCE DRAG, FUEL USE, AND EMISSIONS
• AIR CARGO OPERATORS
• CIVILIAN PASSENGER AIRCRAFT PAIRS
• DISSIMILAR AIRCRAFT PAIRINGS
ex: fighter-tanker missions
• 3-SHIP STAGGERED V FORMATIONS
• 4+ AIRCRAFT FORMATIONS
- string stability- downstream wake effects
• HALE• SMALL UAVS
Q: what is the lower size limit?
Extended Formation Flight ResearchClose Formation Flight Research
2
Prior Wake Surfing Flight ResearchF
ue
l S
avin
gs (
lbs)
Flight Time
Ray, Ronald J., et al, “Flight Test Techniques Used to Evaluate Performance
Benefits During Formation Flight,” 2002Bieniawski, Stefan R., et al, “Summary of Flight Testing and Results
for the Formation Flight for Aerodynamic Benefit Progam,” 2014
3
Wake Surfing Challenges
DR. ERBSCHLOE, USAF AMC CHIEF SCIENTIST (2008):
“WE WILL ONLY BE INTERESTED IN FORMATION FLYING
FOR AERODYNAMIC BENEFIT IF IT IS:
• SAFE.• AIRCREW FRIENDLY.• AIRCRAFT FRIENDLY.• MAKES BUSINESS SENSE.• MAKES OPERATIONAL SENSE.”
OTHER CHALLENGES IDENTIFIED BY THE USAF:
• PILOT TRAINING
• PILOT TACTICAL DUTY DAY RESTRICTIONS
• EQUIPAGE FOR AIRCRAFT OTHER THAN THE C-17• DOMESTIC AND FOREIGN AIR TRAFFIC CONTROL
INDUSTRY PERSPECTIVES, WAKENET USA 2013:• AIR CARGO COMPANIES
• MAJOR CARRIERS AND REGIONAL AIRLINES
• AIRLINE PILOT ASSOCIATION
CHALLENGES IDENTIFIED BY INDUSTRY:
• LACK OF CIVILIAN AIRFRAME DATA
• PASSENGER DISCOMFORT
• WAKE CROSSING PREVENTION
• COST OF EQUIPAGE
• AIR TRAFFIC CONTROL PROCEDURES
• FAA APPROVAL
DR. ERBSCHLOE, USAF AMC CHIEF SCIENTIST (2008):
“WE WILL ONLY BE INTERESTED IN FORMATION FLYING
FOR AERODYNAMIC BENEFIT IF IT IS:
• SAFE.• AIRCREW FRIENDLY.• AIRCRAFT FRIENDLY.• MAKES BUSINESS SENSE.• MAKES OPERATIONAL SENSE.”
OTHER CHALLENGES IDENTIFIED BY THE USAF:
• PILOT TRAINING
• PILOT TACTICAL DUTY DAY RESTRICTIONS
• EQUIPAGE FOR AIRCRAFT OTHER THAN THE C-17• DOMESTIC AND FOREIGN AIR TRAFFIC CONTROL
INDUSTRY PERSPECTIVES, WAKENET USA 2013:• AIR CARGO COMPANIES
• MAJOR CARRIERS AND REGIONAL AIRLINES
• AIRLINE PILOT ASSOCIATION
CHALLENGES IDENTIFIED BY INDUSTRY:
• LACK OF CIVILIAN AIRFRAME DATA
• PASSENGER DISCOMFORT
• WAKE CROSSING PREVENTION
• COST OF EQUIPAGE
• AIR TRAFFIC CONTROL PROCEDURES
• FAA APPROVAL
4
NASA G-III Wake Surfing Flight Experiment
LEAD AIRPLANE:
• NASA G-III
• PRODUCTION AVIONICS WITH
ADS-B OUT
• CABIN VIBRATION SENSORS
TRAIL AIRPLANE:
• NASA C-20A (G-III MILITARY VARIANT)
• PRODUCTION AVIONICS AUGMENTED WITH:
• EXPERIMENTAL PROGRAMMABLE AUTOPILOT
• PILOT TABLET DISPLAYS
• COMMERCIAL ADS-B IN
• VIDEO RECORDING OF FUEL FLOW
• CABIN VIBRATION AND NOISE SENSORS
5
ADS-B Enabled Experimental Autopilot
1090 MHz ADS-B Data Link
• Non-secure data link
• Broadcast twice-per-second at random variations
• Horizontal resolution ~16.7 feet / 1 knot
• Vertical resolution ~25 feet / 64 ft per min
• Accuracy dependent on transmitting avionics
• No wind or weather information
Research Autopilot
• Wake drift and descent predictions
• Wake-relative navigation
• Trajectory control
• Analog ILS localizer and glideslope commands
• Throttle cues to pilot display
Operator Interfaces
• Lead aircraft selection
• Controller gains and parameters
• 3-axis wake-relative position commands
• Arm / engage / disengage
TEST OPERATIONS:
• MACH 0.7, 35,000 FEET
• 4000 FEET IN TRAIL
Experiment Conditions6
TEST CONDITIONS
• DAY VMC
• CALM TO LIGHT TURBULENCE
• 30-40 MINUTE TEST LEGS
• W-291 RESTRICTED AIRSPACE
OVER THE PACIFIC OCEAN
• CONTRAILS PREFERRED BUT
NOT REQUIRED
TEST METHOD:
Experiment Methodology7
10 fps
5 f ps
2 f ps
1 f ps
0.5 f ps
10 fps
5 f ps
2 f ps
1 f ps
0.5 f ps
Reported Lead
Location
Predicted
Wake
Location
wind drift wake
descent
Cross-Track, ft
Vert
ical
Posit
ion
, ft
-100 0 100 200 300 400
-100
-50
0
50
1001. WAKE-FREE TARE
(MINIMUM OF 3 MINUTES)
2. WAKE INGRESS / MAPPING
3. STABILIZED WAKE SURFING:PERFORMANCE DWELL AND
RIDE QUALITY EVALUATION
(MINIMUM OF 5 MINUTES)
1
3
2
4
The autopilot computes a wind- and descent-corrected trajectory for the trail airplane. This trajectory is relative to the lead airplane’s wake.
One knot of error in cross-track wind speed adds 10 ft of error to the predicted wake location.
4. WAKE-FREE TARE
(MINIMUM OF 3 MINUTES)
performance benefits
8
ሶ𝑚0 𝑚 = ሶ𝑚𝑅 𝑚 −𝜕 ሶ𝑚
𝜕𝑉Δ𝑉 −
𝜕 ሶ𝑚
𝜕 ሶ𝑉ሶ𝑉
cockpit fuel flow meters
Fuel Flow Estimation
1 2 3 4 5 6 7
-5
0
5
10
Test Point
Off
-tri
m A
irsp
ee
d,
kca
s
Analysis Point Airspeed, mean
Analysis Point Airspeed, range
-0.1 -0.05 0 0.05 0.1
30
32
34
36
38
Airspeed Rate, kcas/sec
Fue
l F
low
, P
PH
/100
Flight Measurement
Linear Fit
Trim Fuel Flow Estimate
9
Fuel Flow Estimation
8101214161820
32
33
34
35
Fuel Quantity, lb x 1000
Fue
l F
low
, P
PH
/100
Tare Point Trim Estimates
Estimate Uncertainty
Tare Quadratic Fit
-0.2 -0.1 0 0.1 0.2
-8
-4
0
4
8
Off
-Tri
m F
ue
l F
low
, P
PH
/10
0
Airspeed Rate, kcas/sec
Airspeed Rate Correction
Tare Point Data
0 50 100 150 200 250 300 350 400 450
28
30
32
34
36
38
40
Fue
l F
low
, P
PH
/100
0 50 100 150 200 250 300 350 400 450
-0.1
-0.05
0
0.05
0.1
0.15
Air
sp
ee
d R
ate
, kcas/s
ec
Time, sec
Wake-Based Correction
Tare-Based Correction
Mean Tare-Corrected
Measurement
Steady-State Time Segments
𝜕 ሶ𝑚
𝜕 ሶ𝑉for all tare pointsvalue of
tare fuel flow vs. fuel quantity
Example fuel flow corrections for in-wake
performance point.
(local linear fit vs. tare-based correction)
810121416182028
29
30
31
32
33
34
35
Fuel Quantity, lb x 1000
Fue
l F
low
, P
PH
/100
Tare Point Trim Estimates
Estimate Uncertainty
Tare Quadratic Fit
10
Fuel Flow Reduction Results
Seven test points were
completed on the final flight.
810121416182028
29
30
31
32
33
34
35
2%
5%
8%
10% 2%
5%
8%
10%
TP1 TP2 TP3 TP4 TP5 TP6 TP7
Fuel Quantity, lb x 1000
Fue
l F
low
, P
PH
/100
Tare Quadratic Fit
Wake Turbulence Onset
In Wake Effect, 1 Hz Estimate
Wake Ingress, Segment Mean
Wake Surfing, Segment Mean
Wake ingress was stopped when
wake effects (rumbling) were felt
in the cabin. Post-flight analysis
showed this occurred around
3.5% fuel flow reduction.
: Points of constant throttle
setting (20-plus seconds).
This potentially limited the
maximum measured benefit.
Fuel Flow Reduction Results
The steep gradient of wake
effects vs. position prevented the
controller from stabilizing at a
single location within the wake
for extended periods.
Two of the test points (2 and 7)
significantly exceeded the 3.5%
ride quality threshold.
A maximum performance benefit
of >8% was achieved briefly,
consistent with previous wake
surfing results.
11
810121416182028
29
30
31
32
33
34
35
2%
5%
8%
10% 2%
5%
8%
10%
TP1 TP2 TP3 TP4 TP5 TP6 TP7
Fuel Quantity, lb x 1000
Fue
l F
low
, P
PH
/100
Tare Quadratic Fit
Wake Turbulence Onset
In Wake Effect, 1 Hz Estimate
Wake Ingress, Segment Mean
Wake Surfing, Segment Mean
Wake Effect Map
80 100 120 140
-20
-10
0
10
20
8%6%
4%2%
Cross Track, ft
Vert
ical
Tra
ck,
ft
Test Point 2
80 100 120 140
-20
-10
0
10
20
8%
6%
4%2%
Cross Track, ft
Test Point 7
theory 2-4% 4-6% 6-8% 8+ %
An independent measure of the
wake location was unavailable
for verification of the wake
prediction algorithm.
In general, the largest benefits
were measured closest to the
predicted core location.
The gradients of the flight
measurements appear to be
more steep than predicted.
12
Secondary Effects
0 1 2 3 4 5 6 7 8
-0.4
-0.2
0
0.2
Fuel Flow Reduction, %
Devia
tio
n f
rom
tri
m,
deg
Angle of Attack, Theory
Mean Pitch Angle, TP 2
Mean Pitch Angle, TP 7
0 2 4 6 80
0.5
1
1.5
2
2.5
Fuel Flow Reduction, %
Defl
ect
ion
, d
eg
Ailerons
0 2 4 6 80
0.5
1
1.5
2
2.5
Fuel Flow Reduction, %
Spoilers
Mean - TP 2
Mean - TP 7
Combined Peak
Pitch Trim
Wake-induced drag savings are
accompanied by a reduction in trim angle
of attack. The flight-measured change in
pitch trim vs. fuel flow reduction matches
theoretical predictions.
Roll Trim
The wake field produces an asymmetric
lift distribution across the wing, resulting
in increased roll trim with higher fuel
savings. The measured aileron and spoiler
deflections show a correlation with
measured fuel flow reduction.
13
passenger ride quality
Passenger Ride Quality Instrumentation:
• Accelerometers mounted to the seat rails of both airplanes• 3-axis accels sampled at 200 Hz• separate accels for low and high frequency measurements• internal data logging with time stamp
• Sound dosimeter with microphone at approximate passenger ear location• records and logs 1-minute time-average sound levels• 100 Hz to 5 kHz, 40-140 dB
• Pre-flight and post-flight surveys of pilots and research crew
• An additional accelerometer was mounted to the ceiling of the aft baggage compartments of both airplanes to measure tail buffeting
Passenger Ride Quality: Cabin Vibration
14
• Occurred in the strongest part of the wake
• Strong variation with fore-aft cabin location
• Described as “rumbling”, compared to light turbulence or a driving on a washboarded road
Passenger Ride Quality: Cabin Vibration
15
0 200 400 600 800 1000 1200-0.2
0
0.2
Fw
d C
ab
in
Test Point 2 Peak Acceleration, gs
Vertical Lateral Longitudinal
0 200 400 600 800 1000 1200-0.2
0
0.2
Mid
Cab
in
0 200 400 600 800 1000 1200-0.2
0
0.2
Tare Point Wake Ingress Wake Surfing
Time, sec
Aft
Bag
ga
ge
Passenger Ride Quality: Cabin Vibration
16
0 20 40 60 80 100
-50
-40
-30
-20
-10
0
Pow
er/
fre
qu
ency (
dB
/Hz) Fwd Cabin, Lateral
Calm Air
Turbulence
0 20 40 60 80 100
-50
-40
-30
-20
-10
0
Mid Cabin, Lateral
Lead Aircraft
Wake Surfing
Trail Aircraft
0 20 40 60 80 100
-50
-40
-30
-20
-10
0
Frequency (Hz)
Pow
er/
fre
qu
ency (
dB
/Hz) Fwd Cabin, Vertical
0 20 40 60 80 100
-50
-40
-30
-20
-10
0
Frequency (Hz)
Mid Cabin, Vertical
0 20 40 60 80 100
-50
-40
-30
-20
-10
0
Aft Baggage, Lateral
0 20 40 60 80 100
-50
-40
-30
-20
-10
0
Frequency (Hz)
Aft Baggage, Vertical
• Wake-induced vibrations are similar to those of light turbulence at higher frequencies
• Light turbulence contains low frequency content not found in the wake
• Cabin vibrations on the lead airplane during wake surfing were similar to non-turbulent conditions, suggesting measured effects on the trail airplane were due to flight within the wake
Calm Air Wake Ingress Wake Surfing Turbulence 80
81
82
83
84
Ave
rage
No
ise L
eve
l, d
B
Test Point 2
Test Point 7
Passenger Ride Quality: Cabin Noise
17
• A similar increase in noise was also recorded during the more severe of the two “light turbulence” turbulent tare points.
• Slighty increased cabin noise levels were recording during wake surfing, as compared to flight in calm air and in the weaker portions of the wake.
Dosimeter noise recorder,trail aircraft cabin installation
In the 1970s, Jack Leatherwood and others at NASA LaRC conducted a series of studies to develop a criteria to predict passenger discomfort due to vibration and noise.
• Vibration Tests• 2200 test subjects• motion simulator fitted with six tourist-class aircraft seats• 10 - 15 second excitations• lateral, vertical, longitudinal, roll, and pitch vibrations• rated as “comfortable” or “uncomfortable”
• Noise and Vibration Tests• 60 test subjects• combinations of noise and vibration• 4 sound levels, 6 octave bands
NASA Passenger Ride Quality Metric
18
from “Human Discomfort Response to Noise Combined With Vertical Vibration,” Leatherwood, April 1979
NASA Passenger Ride Quality Metric
19
• Frequency-weighted acceleration measurements are combined to form a Discomfort Metric: DISC.
• For sinusoidal vibrations, the DISC metric was developed with the following excitations:
Vertical: 1 - 30 Hz | 0.04 - 0.34 g
Lateral: 1 - 10 Hz | 0.04 - 0.34 g
Roll: 1 - 4 Hz | 0.23 - 2.62 rad/s2
• A DISC of 1 predicts that 50% of passengers will find the ride uncomfortable.
• Note: Leatherwood’s Noise and Duration corrections were not applied for the following results.
from “A Design Tool for Estimating Passenger Ride Discomfort Within Complex Ride Environments,” Leatherwood, Dempsey, and Clevenson, Human Factors, June 1980
the lateral 10-Hz weigting was applied to all data above 10 Hz
0 2 4 6 80
0.5
1
1.5
2
3.5
%
onset established
Fuel Flow Reduction, %
Dis
co
mfo
rt M
etr
ic
Fwd Cabin, onset
Mid Cabin, onset
Fwd Cabin, established
Mid Cabin, established
NASA Passenger Ride Quality Metric
20
0 50 1000
1
2
3
4
Passengers Uncomfortable, %
Dis
co
mfo
rt M
etr
ic
Fwd Cabin
Calm Air
Peak In-Wake
Turbulence
0 50 1000
1
2
3
4
Passengers Uncomfortable, %
Mid Cabin
NASA Criteria
Lead Aircraft
• DISC metric values were calculated for lateral and vertical vibrations recorded at the forward and mid-cabin locations.
• DISC plotted vs. fuel flow reduction shows the gradual onset of wake discomfort below 3.5%.
• Above 3.5% the DISC is consistently high.
• Using the Leatherwood criteria, the peak DISC values calculated at the two cabin locations during wake surfing fall within the region of values measured for light turbulence.
• Even in calm air, the DISC values are quite high, suggesting this metric may over-predict passenger discomfort.
ISO-2631 Passenger Ride Quality Metric
21
• Frequency-weighted acceleration measurements are combined to form a Vibration Dose Value: VDV.
• The ISO metric addresses the following frequency ranges:
0.5 - 80 Hz: health, comfort, and perception
0.1 - 0.5 Hz: motion sickness
• VDV-based human comfort rating predictions increase with exposure time raised to the ¼ power.
• Motion sickness increases with the square root of the exposure time.
• The ISO standard gives a relationship between VDV and descriptive “likely reactions” in terms of comfort value.
ISO-2631 Passenger Ride Quality Metric
22
0 5 100
1
2
3
Exposure, hours
Vib
ratio
n D
ose V
alu
e
Fwd Cabin
Calm Air
Turbulence
Wake Surfing
0 5 100
1
2
3
Exposure, hours
Mid Cabin
0 2 4 6 8 100
5
10
15
Exposure, hours
Moti
on S
ickn
ess
Ra
te,
%
Calm Air
Turbulence
Wake Surfing
• The ISO metric predicts increased passenger discomfort due to wake surfing vs. calm air, although not as severe as light turbulence.
• Even the calm air predictions are solidly ‘uncomfortable’ for flights longer than 5 hours, which may indicate over-prediction of passenger discomfort by this metric.
uncomfortable
veryuncomfortable
extremely uncomfortable
• The ISO metric predicts no appreciable increase in passenger motion sickness due to wake surfing vs. flight in calm air, and significantly less motion sickness than flight through light turbulence.
Summary of the post-flight questionnaires:
• 9 participants (2 pilots, 6 engineers, 1 videographer); majority are frequent flyers
• Wake Surfing Comfort Response:• “Comfortable”: 45% (4 of 9)
• “Neutral”: 45% (4 of 9)
• “Uncomfortable”: 10% (1 of 9)
• 10% reported “Writing” would be difficult
• 33% reported “Sleeping” would be difficult
Comments:
• “Similar to light turbulence”
• “Rhythmic, pulsing sound - not unpleasant but noticeable”
• “Like driving over a slightly-washboarded road”
• “I found the view of contrails outside my window unsettling”
• “The appearance of the wake was larger than I had originally imagined”
Passenger Ride Quality Survey
23
Conclusions:
1. ADS-B is adequate for moderate wake surfing benefits.
2. Accurate wind estimates are critical for wake prediction.
3. Sustained fuel savings are possible above 5% for wake surfing at extended trail distances.
4. There is significant ride quality degradation at higher fuel flow savings.
5. Automatic control is a necessity, including throttles.
Recommendations:
1. Develop and test robust wake estimation, performance optimization, and wake-crossing prevention algorithms.
2. Through modeling and flight research, improve understanding of the causes of ride quality degradation.
3. Characterize wake strength, descent, and decay downstream of the trail airplane.
4. Develop routing and scheduling algorithms for civil operators, and meta-aircraft operations for air traffic control.
Summary
24
Questions?
B-757NASA SUCCESS Mission Video
DC-8NASA ACCESS Mission Video
Examples of Wake Dynamics
B-737© B. Whittaker
B-767A. Brown, AIAA 2007-289
Relative Navigation and Wake Prediction
Wake prediction functions in the autopilot compute a wind-corrected trajectory for the trail airplane. This trajectory is relative to the lead airplane’s wake.
Timing uncertainty in ADS-B messages results in larger errors in along-track vs. cross-track.
The trail airplane flies a wake-relative trajectory.
-50 0 50-250
-200
-150
-100
-50
0
50
100
150
200
Cross-Track Error, ft
Alo
ng
-Tra
ck E
rro
r, f
t
1-
2-
4%6%8%
Drag Reduction Error Budget
One knot of error in cross-track wind speed adds 10 ft of error to the predicted wake location.
N
Simplified Wake Location Prediction
ξ
ψ
Gt
Xtrack
Ltrackξ
ξ = ψ - Gt
+
+
+
+
++
+
+
+
+
+
+
+
+
++
+
+
Parallel ground tracks
Estimated Vortex
Formation Axis
N
ξ
WE
Despite good results in the
piloted sim, the pilots initially
found the throttle cues
“Unsatisfactory” in flight.
For the final flight, the pilot
along-track error cue was re-
designed with an increased
range of view, and a relaxed
acceptable error criteria.0 0.05 0.1 0.15 0.2
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Duty Cycle
Agg
ress
iven
ess
ACT Pilot Throttle Workload Metric
Config A Tare
Config A Wake
Config B Tare
Config B Wake
The modified display reduced the pilot
workload to “Satisfactory” and improved
post-flight calculation of fuel flow savings.
Pilot Throttle Cue and Wake Display