You are on page 1of 12

Downloaded from SAE International by Univ of California Berkeley, Monday, August 06, 2018

SAE TECHNICAL
PAPER SERIES 2002-01-2912

Wind-Tunnel Investigation of Commercial


Transport Aircraft Aerodynamics at
Extreme Flight Conditions
Gautam H. Shah, Kevin Cunningham, John V. Foster,
C. Michael Fremaux and Eric C. Stewart
NASA Langley Research Center

James E. Wilborn, William Gato and Derek W. Pratt


The Boeing Company

World Aviation Congress & Display


Phoenix, Arizonia
November 5-7, 2002

400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A. Tel: (724) 776-4841 Fax: (724) 776-5760 Web: www.sae.org
Downloaded from SAE International by Univ of California Berkeley, Monday, August 06, 2018

For permission and licensing requests contact:

SAE Permissions
400 Commonwealth Drive
Warrendale, PA 15096-0001-USA
Email: permissions@sae.org
Fax: 724-772-4028
Tel: 724-772-4891

For multiple print copies contact:

SAE Customer Service


Tel: 877-606-7323 (inside USA and Canada)
Tel: 724-776-4970 (outside USA)
Fax: 724-776-1615
Email: CustomerService@sae.org

ISSN 0148-7191

Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE. The author is solely
responsible for the content of the paper. A process is available by which discussions will be printed with the paper if it is published in
SAE Transactions.

Persons wishing to submit papers to be considered for presentation or publication through SAE should send the manuscript or a 300
word abstract to Secretary, Engineering Meetings Board, SAE.

Printed in USA
Downloaded from SAE International by Univ of California Berkeley, Monday, August 06, 2018

2002-01-2912

Wind-Tunnel Investigation of Commercial Transport Aircraft


Aerodynamics at Extreme Flight Conditions
Gautam H. Shah, Kevin Cunningham, John V. Foster, C. Michael Fremaux and Eric C. Stewart
NASA Langley Research Center

James E. Wilborn, William Gato and Derek W. Pratt


The Boeing Company

ABSTRACT effort involves acquiring an extensive, detailed


aerodynamic database for determining flight
A series of low-speed static and dynamic wind tunnel characteristics not only beyond the conventional flight
tests of a commercial transport configuration over an envelope, but also beyond the normal experimental,
extended angle of attack/sideslip envelope was wind-tunnel-derived database typically acquired by a
conducted at NASA Langley Research Center. The test manufacturer. Methods to validate the resultant
results are intended for use in the development of an database to increase confidence in providing accurate
aerodynamic simulation database for determining aircraft simulation of extreme flight conditions are also being
flight characteristics at extreme and loss-of-control developed within this effort.
conditions. This database will be used for the
development of loss-of-control prevention or mitigation A further goal of the CUPR effort is to assess the
systems, pilot training for recovery from such conditions, accuracy and utility of existing aerodynamic databases
and accident investigations. An overview of the wind- at extreme angles of attack and sideslip, for both static
tunnel tests is presented and the results of the tests are and dynamic characteristics. For this flight region data
evaluated with respect to traditional simulation database are often either estimated or extrapolated from lower α/β
development techniques for modeling extreme wind tunnel results because there has been no previous
conditions to identify regions where simulation fidelity requirement for high fidelity in this regime, and due to
should be addressed. the difficulty in safely validating such characteristics with
full-scale flight tests.
INTRODUCTION
Recent accidents have demonstrated that aircraft can
Loss-of-control has been identified as a major readily reach flight conditions beyond the range of a
contributor to the fatal accident rate of large, transport- manufacturer’s existing aerodynamic database (Figure
category airplanes. Loss-of-control accidents involve 1), and current extrapolations from that database may be
extremely large angles of attack and sideslip, as well as insufficient to properly model the aircraft’s response for
large angular rates, all of which are far outside the simulation training of recovery from such conditions. In
normal operational envelope. Recently, it has been addition, dynamic aerodynamic characteristics are
recognized that improved aerodynamic models of these usually estimated, rather than measured, for an even
extreme conditions, which can be generated by control more restricted envelope of confidence than that for the
upsets, unusual attitudes, atmospheric disturbances, static characteristics.
loss-of-control and/or out-of-control situations, are
needed to develop systems that prevent/mitigate upset To acquire an appropriately accurate and detailed
conditions and to facilitate pilot training of techniques for aerodynamic database for modeling of extreme
recovery from these various situations1,2. conditions, an unprecedented series of static and
dynamic low-speed wind tunnel tests was conducted in
As a part of the NASA Aviation Safety Program (AvSP), 2000 and 2001 on a conventional twin-engine
a research effort has been undertaken under the Control commercial transport configuration in facilities at NASA
Upset Prevention and Recovery (CUPR) sub-element to Langley Research Center. An overview of the initial
improve aerodynamic modeling and simulation of results of those tests is presented in this paper.
commercial transport aircraft at flight attitudes and
conditions beyond the normal operating envelope. This
Downloaded from SAE International by Univ of California Berkeley, Monday, August 06, 2018

TEST DESCRIPTION lower surfaces of the wing, stabilizer, and vertical tail.
Comparisons of test results were made to previous,
Tests were conducted in the NASA Langley 14x22-Foot higher Reynolds Number wind-tunnel tests conducted on
Tunnel and 20-Foot Vertical Spin Tunnel (VST) using a same-size model of a similar vehicle configuration.
two different sized models of a conventional, medium- Trends and differences due to Reynolds Number in the
range twin-engine commercial transport configuration. low angle-of-attack range were as expected, and
Conventional static and forced oscillation (about all three became less significant at larger angles of attack and
axes) tests were conducted on a 5.5%-scale model, and sideslip and fully separated flow conditions (also as
rotary balance tests were conducted on a 3.5%-scale expected).
model. Model geometry information and control surface
locations, as well as installation photographs, are NOMENCLATURE
presented in Figure 2.
Longitudinal data are referenced to stability axes, and
The data coverage was extremely broad in many lateral/directional data are referenced to body axes.
aspects, with large angle of attack and sideslip ranges, Moment coefficients are referenced to 25% Mean
and control surface deflections both within and beyond Aerodynamic Chord ( c ). Variables presented or
the normal range of motion. The aerodynamic discussed are:
contributions of individual components, such as vertical
fin, horizontal stabilizer, or engine/pylon were also b Wing span, ft
studied. Data were acquired for both a clean-wing, c Mean aerodynamic chord, ft
flaps/slats-retracted configuration as well as a high-lift CL Lift coefficient, stability axis
configuration with both flaps and slats extended. The Cm Pitching moment coefficient, stability axis
results presented in this paper are for the clean-wing Cl Rolling moment coefficient, body axis
configuration. Clβ̇ Rolling moment coefficient due to sideslip rate,
body axis
Conventional static tests were conducted in the 14x22- Clp̂ Rolling moment coefficient due to non-dimensional
Foot Tunnel at a dynamic pressure of 10 lb/ft2, roll rate, body axis
corresponding to a Reynolds Number of 0.54x106 based Clp Roll damping coefficient, Clpˆ + Clβ˙ sin α , body axis
on mean aerodynamic chord. Data were acquired at Cn Yawing moment coefficient, body axis
angles of attack from –30° to +90°, and at angles of Cnβ̇ Yawing moment coefficient due to sideslip rate,
sideslip from –45° to +45°. A limited amount of testing body axis
was also conducted at a dynamic pressure of 30 lb/ft2 to Cnr̂ Yawing moment coefficient due to non-dimensional
explore Reynolds Number effects. yaw rate, body axis
Cnr Yaw damping coefficient, Cnrˆ - Cnβ˙ cos α , body axis
Forced oscillation tests were also conducted in this
p Roll rate, body axis, rad/sec
facility. The model was oscillated in a sinusoidal motion
p̂ Non-dimensional roll rate, pb/2V∞, body axis
over a range of frequencies and amplitudes that
corresponded to typical full-scale short-period and r Yaw rate, body axis, rad/sec
Dutch-roll motions. Oscillation tests were conducted r̂ Non-dimensional yaw rate, rb/2V∞,body axis
individually in the pitch, roll, and yaw body axes to S Wing reference area, ft2
acquire aerodynamic damping characteristics. V∞ Freestream velocity, ft/sec
α Angle of attack, deg
Dynamic tests were also conducted in the 20-Foot β Angle of sideslip, deg
Vertical Spin Tunnel on a conventional rotary balance β̇ Sideslip rate, rad/sec
apparatus to acquire steady-state rotary damping ∆Cl Incremental pitching moment coefficient due to
characteristics. During rotary balance tests, the model is aileron deflection
rotated at both positive and negative rates in order to ∆Cm Incremental pitching moment coefficient due to
capture the potentially significantly non-linear nature of elevator deflection
the data, especially at non-zero sideslip angles. Data ∆Cn Incremental yawing moment coefficient due to
were acquired at angles of attack from 0° to +90°, and at rudder deflection
angles of sideslip from –45° to +45°. Ω Spin rate, rad/sec

For all tests, static and dynamic, six-component force Control surface sign conventions are as follows:
and moment data were acquired using an internal strain-
gauge balance. Aileron: Positive Trailing Edge Down
Stabilizer: Positive Trailing Edge Down
To minimize Reynolds Number effects, grit strips were Elevator: Positive Trailing Edge Down
used to fix transition at 10% chord on the upper and Rudder: Positive Trailing Edge Left
Downloaded from SAE International by Univ of California Berkeley, Monday, August 06, 2018

DISCUSSION OF RESULTS discontinuities, although reduced stability (pitch-up) at


high-α and high-β is noted.
Because of the proprietary nature of the basic results,
absolute numerical values cannot be presented in most For some aircraft simulation databases, the
cases. Hence numerical values are absent on the aerodynamics of the full aircraft configuration are
vertical axes of the data charts; however the data are modeled as the sum of several components, such as
presented in conventional form, with values typical to wing/body and a stabilizer increment, as opposed to
those of conventional transport aircraft, and are intended direct incorporation of full-configuration data. In such
to be used to provide general trend information for instances it becomes critical to accurately model the
applicability to many transport-category aircraft. pitch stability variations of each component. Minor over-
predictions of stability due to extrapolations can add up
SIMULATION EXTRAPOLATIONS to a significant over-prediction of the complete aircraft’s
stability.
‘Nominal Simulation Extrapolation’ (NSE) lines are
shown on some of the data plots, and are intended to This ‘build-up’ method is illustrated by example in figure
reflect traditional simulation implementation practices for 3e, which shows pitching moment for the wing/body
transport-category aircraft, in which aerodynamic alone, for an increment due to the stabilizer (calculated
characteristics are held constant for angles of attack and as the difference between wing/body and full-aircraft
sideslip beyond the range of a typical wind-tunnel configurations), and for the full-aircraft configuration.
database. The NSE lines shown in this paper are based The NSE for the full configuration is the sum total of the
on holding aerodynamic values constant for angles of NSE’s for the wing/body and stabilizer. As the figure
attack and sideslip in excess of 30° and ±30°, shows, the over-prediction in pitch stability (for 22° < α <
respectively. This database boundary closely 30°) for the stabilizer increment extrapolation results in a
approximates that used for an aerodynamic simulation similar over-prediction for the full configuration. While in
database of a commercial transport aircraft configuration this example the over-prediction of stability is small, it
similar to this research configuration. should be recognized that superimposition of multiple
extrapolations can produce large errors in predicted
LONGITUDINAL AERODYNAMICS AND STABILITY stability, as seen for α > 50°. Furthermore, in situations
where the stabilizer (or other surface) increment is
The basic static longitudinal aerodynamic and stability independently derived through analytical or other means,
characteristics are shown in figures 3a-b, as a function extrapolations of their estimated value over large angle
of stabilizer deflection, with all other controls at zero of attack ranges can result in substantial over- or under-
deflection. Although the initial stall break occurs just predictions of an aircraft’s stability. The implications of
above α = 10°, increases in lift are seen up to α = 35°, this will be further discussed in the longitudinal control
along with substantial drag increments (not shown). The power section.
pitching moment remains generally stable up to stall
angle of attack for nominal stabilizer deflections. LATERAL/DIRECTIONAL AERODYNAMICS AND
Changes and variations in pitch stability to neutral or STABILITY
even slightly unstable are seen across the angle of
attack range when the stabilizer is beyond its local stall The basic directional aerodynamic and stability
angle of attack. characteristics are shown in figures 4a-b. Directional
stability degrades above α = 15°, due to increasing
Normal stabilizer travel for this configuration is shielding of the vertical tail by the fuselage, and
approximately +4° to –12°, providing an angle-of-attack diminishes to negligible levels by α = 30° for sideslip up
trim range of about –2° to +23° (for a center-of-gravity to 20° (Figure 4a). Above α = 55°, the generally highly
location at 0.25 c ) while allowing full elevator use for separated flowfield results in extremely non-linear and
control. However, an additional 4° of stabilizer travel to erratic behavior. The significantly higher stability at β =
+8° continued to provide a linear increase in nose-down 30° is assumed to be due to the vertical tail moving out
moment, which could be beneficial in compensating for a of the fuselage wake. Stability is fairly linear at low-
failed or jammed elevator system. A doubling of normal α/low-β, and diminishes at high sideslip (Figure 4b). At
stabilizer travel in that direction is admittedly large, high sideslip, the stability characteristics appear to
however, it can be assumed that proportional gains may collapse into three general areas: pre-stall (α < 10°), stall
be realized with more modest increases in stabilizer (10° < α < 20°), and post-stall (α > 20°). A directional
travel range. instability is seen at sideslips of 15°-25° at α = 40° and
60°.
The behavior of the longitudinal characteristics with
sideslip is shown in figures 3c-d. Out to β = 45°, the
Lateral stability (Figures 4c-d) sharply diminishes above
characteristics are fairly well behaved, with no sharp α = 12°, coinciding with wing stall, and is essentially
Downloaded from SAE International by Univ of California Berkeley, Monday, August 06, 2018

neutral at angles of attack of 20°-30° for small sideslip can therefore substantially over-predict available control
(Figure 4c). At higher angles of attack and sideslip, power during flight at stall and post-stall conditions.
stability levels return. At low angles of attack, the
flattening and reversal of the curves at high sideslip For low angle of attack, control power is retained up to
indicate that while still remaining stable overall, local very high levels of sideslip in the direction where rudder
instabilities may be present at such conditions (Figure deflection would command the sideslip (i.e., positive
4d). sideslip/positive rudder or negative sideslip/negative
rudder) (Figure 6b). However, when rudder deflection is
However, because wind tunnel databases typically hold used to oppose the sideslip (i.e., negative
constant values beyond approximately 30° in angle of sideslip/positive rudder or positive sideslip/negative
attack or sideslip, these substantial stability variations rudder), the control power can be reduced significantly at
(and subsequent post-stall roll/yaw responses) may not higher sideslips due to stalling of the vertical tail surface
be appropriately modeled in current transport simulators. at large rudder deflections. Whereas the aircraft’s basic
directional stability would provide recovery from a
LONGITUDINAL CONTROL POWER transient disturbance, this loss of yaw control power in
sideslip could become a significant factor in situations
Pitch control power with zero stabilizer is shown in where an external influence (e.g., engine-out) requires
figures 5a-b. The total range of travel (50°) is typical for the aircraft to fly in a stabilized sideslip, with the resultant
this configuration. Elevator power is relatively linear up diminished control power reducing maneuvering
to about ±20° of deflection at low angles of attack. capability. As seen with respect to angle of attack,
Incremental pitching moment, with nominal simulation simulation extrapolations also tend to over-predict
extrapolation, is shown in figure 5c, and indicates a available control power at extreme sideslip angles.
progressive degradation in control power from wing stall
up to α = 90°. Roll control power due to a single deflected (right)
aileron is shown in figures 7a-b. Because of the small
Current simulations generally adequately model magnitude of rolling moment, the vertical axis is
transport aircraft aerodynamic characteristics for angles magnified to twice that shown in figures 4c-d. A sharp
of attack up to stall. However, as indicated in Figure 5c, degradation in aileron power is evident above α = 10°,
simulation extrapolation assumes a constant level of coinciding with wing stall (Figure 7a). Between stall and
control power beyond these angles of attack, resulting in α = 40°, aileron power remains essentially constant, and
a potentially significant over-prediction of available pitch then diminishes at higher angles of attack. Again,
control power in this flight regime. Coupled with the simulator extrapolation tends to over-predict available
potential for over-prediction of high-alpha pitch stability control power at high angles of attack. In sideslip
discussed earlier, existing transport aircraft simulators (Figure 7b), roll control power from the leeward wing
can present post-stall penetration/departure diminishes as sideslip increases. Because control
characteristics that are substantially more benign than power is fairly constant with beta at high sideslips, the
those that would actually be experienced in flight, with flat-line simulation extrapolations do not cause a large
greater control authority and a far stronger tendency for adverse impact on control power predictions.
recovery predicted by simulation. Because stall
recovery procedures and altitude loss predictions are LATERAL/DIRECTIONAL DAMPING
based on such simulations it becomes extremely CHARACTERISTICS
important that longitudinal stability and control power
characteristics be accurately reflected in the Roll and yaw damping characteristics from forced-
aerodynamic databases. oscillation tests are presented in figures 8a-b, along with
nominal simulation extrapolations for comparison.
LATERAL/DIRECTIONAL CONTROL POWER
Damping characteristics for transport category aircraft
Rudder control power is presented in figures 6a-b. The are generally derived by the manufacturer through
nominal deflection range of the rudder for this analytical or empirical means, as opposed to
configuration is approximately ± 30°; only a small experimental ground tests, and are validated with flight
additional benefit in control power is seen with additional data at low angles of attack. They are typically modeled
deflection up to 45° (Figure 6a). Yaw control begins to as near-constant values throughout the angle of attack
diminish above α = 15°, and is virtually negligible above range.
α = 55°. As with the longitudinal characteristics,
however, simulation extrapolation at high angles of The NSE’s in Figures 8a-b reflect that approach,
attack, being held constant above α = 30°, may not showing some variation at low alpha, but are then held
adequately account for the full level of degradation and constant starting at angle of attack just above stall. The
non-linearities and degradation in stability (particularly in
Downloaded from SAE International by Univ of California Berkeley, Monday, August 06, 2018

the roll axis) below and through stall angle of attack are fidelity aerodynamic databases for flight in extreme
not generally well-modeled in simulation. conditions.

The diminished roll damping and highly non-linear yaw Current simulation database extrapolations to such
damping at stall and post-stall angles of attack can result extreme conditions generally do not model the
in larger bank angle and sideslip excursions during substantial nonlinearities, instabilities, and control
maneuvering at those flight conditions than would be authority degradations that are present at those
predicted in a simulation using the current database conditions, and, as a result, can predict flight
modeling methods. These differences between wind- characteristics that are much more benign than would be
tunnel and existing-database damping characteristics, experienced in actual flight. In addition, because the
along with the previously-mentioned differences in static extrapolations tend to hold constant values throughout
lateral stability characteristics, can result in stall and stall high alpha/beta conditions, the resulting response does
recovery dynamics that are substantially different from not realistically vary through the post-stall regime.
predictions.
Because it is assumed that aerodynamic characteristics
ROTARY CHARACTERISTICS at such conditions, where the degree of flow separation
is high, are relatively insensitive to minor (or even
A representative plot of the rotary balance data is moderate) configuration differences, it is expected that
presented in figure 9. In this example, the body-axis the data trends from these and future tests can have
yawing moment coefficient (Cn) is plotted against non- applicability to a wide range of transport aircraft.
 
dimensional spin rate  Ωb  .
 2V∞  REFERENCES
For clarity, only a subset of the angles of attack tested is
shown in the figure. 1. Russell, P.D.; Pardee, J.J.; et al: "Joint Safety
Analysis Team – Loss of Control: Results and
Examination of the figure gives an indication of the Analysis Final Report". Sponsored and published by
predicted damping (or propelling) characteristics of the Commercial Aviation Safety Team, December 2000.
configuration for the conditions of the test. Quadrants II
2. "Airplane Upset Recovery Training Aid". Sponsored
and IV represent damped conditions in that the yawing
and published through a collaborative effort of the
moment is acting in a direction opposite to the rotation
commercial aviation industry, August 1998.
rate (e.g. negative yawing moment with a positive
rotation rate, and vice versa). Conversely, quadrants I
and III represent propelling conditions in that the yawing
moment and rotation rate have the same sign. Although
not shown here, data for the rolling moment and pitching
moment coefficients would be analyzed in a similar
manner. At very high angles of attack (i.e., approaching
90 degrees), yawing moment is the primary driver for a
spin and thus an indication of a propelling condition may
indicate that a spin mode is possible. Figure 9 indicates
that the basic configuration remains damped in yaw over
the entire angle of attack range tested.

CONCLUDING REMARKS

While the aerodynamic stability and control


characteristics of transport-category aircraft are well
known and predictable in the normal operating flight
regime, the body of knowledge of the characteristics
outside of that conventional range, at large angles of
attack and sideslip, is somewhat limited.

Because of the difficulties and high risks involved in


acquiring actual flight data at such conditions, detailed Figure 1. Typical angle of attack/sideslip envelope for
ground test data from wind-tunnel testing are required to commercial transport aircraft
model the aerodynamics in those regimes. The results
of such tests, as described in this paper, are designed to
provide a starting point for developing detailed, high-
Downloaded from SAE International by Univ of California Berkeley, Monday, August 06, 2018

Figure 2b. Static Force Test Set-up, LaRC 14x22

5.5% Model 3.5% Model


NASA Langley NASA Langley
Test Facility
14x22 Tunnel Vertical Spin Tunnel
•Static •Static
Test Techniques
•Forced Oscillation •Rotary Balance
S, ft2 5.902 2.390
b, ft 6.849 4.358
c , ft 0.915 0.582 Figure 2c. Forced Oscillation Test Set-up, LaRC 14x22
Length L, ft 8.539 5.434

Figure 2a. Model Geometry

Figure 2d. Rotary Balance Test Set-up, LaRC VST


Downloaded from SAE International by Univ of California Berkeley, Monday, August 06, 2018

Figure 3a. Basic longitudinal characteristics and Figure 3d. Basic longitudinal characteristics in sideslip;
stabilizer effect; all other controls zero -- Lift all controls zero -- Pitch

Figure 3b. Basic longitudinal characteristics and Figure 3e. Example of ‘build-up’ method for modeling
stabilizer effect; all other controls zero -- Pitch pitch stability characteristics

Figure 3c. Basic longitudinal characteristics in sideslip; Figure 4a. Basic directional characteristics;
all controls zero -- Lift all controls zero
Downloaded from SAE International by Univ of California Berkeley, Monday, August 06, 2018

Figure 4b. Basic directional characteristics; Figure 5a. Elevator effect; all other controls zero – Lift
all controls zero

Figure 4c. Basic lateral characteristics; all controls zero Figure 5b. Elevator effect; all other controls zero – Pitch

Figure 5c. Incremental elevator control power;


Figure 4d. Basic lateral characteristics; all controls zero all other controls zero
Downloaded from SAE International by Univ of California Berkeley, Monday, August 06, 2018

Figure 6a. Incremental rudder control power; Figure 7b. Incremental aileron control power, sideslip
all other controls zero; β = 0° effect at α = 0°; all other controls zero
(vertical scale magnified 2x from figures 4c-d)

Figure 6b. Incremental rudder control power, sideslip Figure 8a. Roll damping characteristics; all controls zero
effect at α = 0°; all other controls zero

Figure 7a. Incremental aileron control power; all other Figure 8b. Yaw damping characteristics; all controls zero
controls zero; β = 0°
(vertical scale magnified 2x from figures 4c-d)
Downloaded from SAE International by Univ of California Berkeley, Monday, August 06, 2018

Figure 9. Rotary yaw damping characteristics, β = 0°;


all controls zero

You might also like