Document created: 29 September 03
Air University Review, May-June 1974
Robert J.
Woodcock
Thomas J. Cord
The early glider flights of the Wright brothers often ended by dropping off on one wing, out of control, with a wingtip eventually striking the Kitty Hawk, North Carolina, sand in a rotary motion. While the low altitude of these flights prevented motion from developing fully, it seems clear that these were departures into incipient spins.
In those earliest days of manned flight, the spin was as dangerous as it is today. When the Wright brothers first tried warping the wings to roll into a turn, they found that the banking was accompanied by a dangerous tendency to diverge in yaw at high angle of attack.1 Adding a fixed vertical fin helped stabilize the 1902 glider, but the loss-of-control problem persisted. Orville Wright reasoned that a hinged vertical rudder could produce a counter yawing moment to keep the yaw from starting and thus enable the flyer to retain control. This was tried first with rudder deflection connected to the wing-warp control, then with the pilot controlling the rudder separately. The fix was effective but required the pilot's constant attention. Proper spin recovery controls were not generally known until 1916, when F.W. Gooden, a British major, conducted flight test experiments on spin recovery procedures in a British F. E. 8.2 For early airplanes the spin recovery technique was at least rational if not instinctive: forward stick and rudder opposing the yawing motion should stop the rotation and unstall the wing. With these recovery controls known, the spin was used as a maneuver to lose altitude without gaining airspeed.3 Then in the mid-1920s, some of the more peculiar spin modes were recognized as problems. Accident summaries from that era4, 5, 6 show spins involved in about three percent of all accidents reported and in twenty to thirty percent of the fatal accidents.
Analytical studies and dynamic wind-tunnel testing to reduce the stall/spin problem were reported as early as 1919. Auto-rotation was observed in the wind tunnels, and the first analytical prediction methods were developed by Glauert. About 1930, a method of determining the flight path and altitude of a spinning aircraft was put into use.7 Rotation rates about and accelerations along the principal axes, as well as vertical velocity, were measured and recorded photographically. This information was used to define the motion of the aircraft, which could then be used in conjunction with the analytical prediction methods.
In the 1920s and 1930s several forms of testing were being performed. Because of the hazards involved in stall/spin flight testing, researchers were hesitant to use full-scale aircraft. One safety measure used in full-scale testing was the attaching of external ballast, which when released would cause the center of gravity of the airplane to move forward, thus returning the airplane to a controllable configuration.8 In general, however, models were used for spin testing. One of the early spin models was dropped from the top of a 100-foot balloon hangar at Langley Field, Virginia. This proved an inadequate means of obtaining data, and soon vertical wind tunnels were being built to investigate spinning (1930 in the United States, 1931 in England). In 1945 the Army Air Force dropped an instrumented model from a Navy blimp to study spin entry and recovery.
The stability and control of airplanes have remained important considerations as aircraft have advanced since the Wright brothers' flights. A degree of both qualities is needed for safe flight, and further stability and control requirements must be met in order for a pilot to perform assigned missions effectively. One critical region of flight is at high angles of attack, where the airplane is susceptible to stalling and possible spinning. Despite the long standing of stall/spin problems, loss of control at high angle of attack is a major factor in the accident rates of our current fighter aircraft such as the F-4 and F-111.
As jet aircraft were developed, the inertial characteristics of fighters in particular were changed to the point that spins and other post-stall motions became more troublesome and even required different recovery techniques. By the time of the 1957 Wright Air Development Center Spin Symposium,9 most stall/spin problems were identified, some analysis methods had been developed, and the electronic digital computer provided a useful tool with which to examine the stall/spin problem.10
Then suddenly the emphasis was shifted to space. With little management interest and rather poor expectations of improvement, resources for stall/spin research were quite limited. Instead the Air Force tended to concentrate on performance improvements, which often have aggravated stability and control problems at high angles of attack. Today, a large and costly Air Force accident record and a renewed emphasis on maneuver capability have led to a larger concentrated effort to solve the problems associated with aircraft operating in the stall/spin flight regime.11
Large aircraft have also experienced stall/spin problems. For example, several B-58s were lost in spins. Automatic trimming of the control-stick force was mechanized in such an insidious way that an inattentive pilot might not be aware of a slowdown to stall speed. Trouble with fuel management could result in an extreme aft center of gravity, at which B-58 stability and control were deteriorated. On long flights the C-133 would climb to an altitude approaching its absolute ceiling. Poor stall warning and a vicious stall while trying to fly there are thought to have caused the disappearance of several C-133 aircraft. It has become customary to require analysis and spin tunnel testing of all military airplanes even though flight demonstration of large, low-maneuverability types is limited to stalls with only moderate control abuse.12
Systematic design data for high angle of attack do not exist; in fact, even aerodynamic force and moment data on specific configurations of current interest are sparse. And there is only a limited degree of confidence. The problems are highly nonlinear in nature, and details vary extensively with aircraft configuration. Since spins and poststall gyrations are no longer useful maneuvers but are to be avoided, there is a great tendency to forget about them unless and until frequent incidents occur. Because of the danger, spins generally are not tested intentionally in flight until well into the production run of an airplane. By then, changes have become very costly to make.
As with most technical areas, stall/spin technology has a particular set of terms that must be precisely defined before the subject can be clearly understood. The following definitions will hold for our purposes.13
stall: The peaking of aerodynamic lift, occurrence of uncommanded aircraft motion about any axis, or onset of intolerable buffet or structural vibration, due to airflow separation induced by high angle of attack (α). The least angle at which one of these phenomena occurs is the stall angle of attack.
post-stall gyration: The uncontrolled motions about one or more airplane axes following departure from controlled flight. This motion normally occurs at and above the stall angle of attack, though lower angles may be encountered intermittently.
spin: A motion characterized by a sustained yaw rotation, with α greater than the stall angle. The spin may be erect or inverted, flat (very high α, 70 to 90 degrees) or steep, and may have oscillations superimposed on the rotary motion.
The military specification14 for flying qualities defines good high-α characteristics in terms that are qualitative rather than quantitative. The airplane must exhibit adequate stall warning, and in addition the stall must be easily recoverable. We require resistance to violent departures from controlled flight, which might induce post-stall gyrations or spins. There are also requirements for recovery from attainable post-stall motions. The definitions of good high-angle-of-attack characteristics will differ for the various classes of aircraft; but with respect to fighter aircraft, a pilot should not have to worry about loss of control while flying within his useful maneuver envelope. Current research will lead, we hope, to quantitative requirements that will be of more use in the design stage for all classes of airplanes.
Generally post-stall design and testing have emphasized spins and spin recovery, taking the point of view that assurance of recoverability from the worst possible out-of-control situation guarantees safety. This philosophy falls short in several respects. Resistance to departure has not been emphasized adequately. The motions can be disorienting, and recovery control inputs like ailerons with the spin are unnatural. And as airplanes grow larger and heavier altitude loss becomes excessive. F-111 instructions, for example, are to eject if spin recovery has not commenced upon reaching 15,000 feet altitude. Spins and spin recovery should not be neglected, but emphasis needs to shift to departure resistance and early recovery.
Flow separation is the common cause of departure. A sharp, highly swept wing leading edge is conducive to leading-edge separation, while trailing-edge separation is typical of a wing having a blunt leading edge with little sweep. Unsteady flow effects are poorly understood for the three-dimensional case of interest for real airplanes. Vortices shed from slender, pointed nose or from the wing-fuselage juncture, for example, can deteriorate flow over the tail, affecting both stability and control. Separated flow over ailerons can destroy their effectiveness. At high angle of attack, deflection of roll control often produces large yawing moments, too. Asymmetric moments may also result from very small configuration asymmetries. A typical fighter has low rolling inertia, which tends to force rolling to be about the x body axis rather than the flight path; this characteristic tends to convert angle of attack into sideslip (β) cyclically as the airplane rolls. But other inertial factors have a stabilizing tendency: some departures occur near the angle of attack at which the effect of the factor expressed by the formula
Cnβ, dyn = Cnβ cos α - Iz C1β sin
α
──
Ix
becomes zero. The aerodynamic rolling-moment derivative C1β, being normally of different sign, will tend to augment the static directional stability, Cnβ. Ix and Iz are the moments of inertia about the airplane roll and yaw axes, respectively; α is the angle of attack.
Recent studies of current Air Force fighters have developed additional insight into their high-angle-of-attack problems. Also, initial F-14 and F-15 flight results, show that substantial improvements can be secured by concentrated design attention, but neither of these airplanes has yet been fully evaluated. In any case, the design effort is very large, trade-offs are uncertain, and confidence is unsure until after thorough flight demonstration. There remains the need to establish definitive requirements and develop a greatly expanded basis for aerodynamic and flight control design. To that end, improved design methods and criteria are being sought for the high-α characteristics of present and future aircraft. The following is a short description of various areas of technology being developed to minimize the high-angle-of-attack problem. More detailed information may be obtained from the document listed in note 11.
Except for a few rules of thumb, analytical stall/spin work is based on the mathematical equations that describe the motion of an aircraft. The general equations of motion involve both inertial and aerodynamic nonlinearities. For the large-amplitude motions associated with stall/ post-stall flight, linearization of the equations is of limited value, while using the nonlinear equations makes generalization difficult. Some attempts are being made to rewrite linear equations by changing the reference axes,15 to give perturbations about a steady spin,16 or to change the form of the aerodynamic description of the aircraft. But in most cases the nonlinear equations are used in order to describe the motion adequately. The primary reason for inaccurate prediction of high-angle-of-attack characteristics is the quality of the available aerodynamic data.
For most configurations, static wind tunnel data are scarce at high angles of attack. Dynamic data are even harder to obtain. For such data as are available, after corrections for tunnel wall effects, tunnel blockage, etc., the remaining effect of disparity of Reynolds number (an indicator of viscous flow effects) can be estimated only uncertainly. We still do not understand some of the phenomena well enough to predict their severity or even their occurrence. In order to estimate aerodynamic forces and moments, it is necessary to predict flow separation lines and pressure distributions in three-dimensional subsonic flow. At this time, more often than not the high-α data must be fudged in order to get analytical predictions to agree with flight-time histories.
The principal U.S. source of high-α static aerodynamic data at high Reynolds number has been the National Aeronautics and Space Administration Ames 11-foot and 12-foot wind tunnels. Although flow irregularities and mount limitations there have limited the validity of data taken, these deficiencies are to be rectified. The Arnold Engineering Development Center 16-foot transonic tunnel also has a high Reynolds number capability, but no existing facility can reach the flight Reynolds numbers of full-scale airplanes. Still, exceeding a critical Reynolds number appears to be the most test requirement. NASA has had a small rotary balance in the spin tunnel and now is employing a new rotary balance in the Langley 30-foot by 60-foot tunnel. Rotary and oscillatory dynamic are expected to produce different aerodynamic results. NASA Ames is also working on improved dynamic wind-tunnel testing apparatus.
The steady spin and recovery can be investigated in a vertical wind tunnel, but the dynamically scaled spin-tunnel aircraft models are rather small. Moreover, a more fundamental limitation is that the models are tossed into the tunnel much as a Frisbee is launched; thus little can be learned about the prespin phase of stalled flight. Also at NASA Langley, models of about 3-foot span are flown in the 30-foot by 60-foot tunnel, with power and control signals transmitted through an umbilical cord. This method enables l-g stalls and initial departures to be studied in conditions resembling free flight.
Free-flight model testing has also taken the forms of catapulted models (recorded by multiple-exposure photos) and radio-controlled models dropped from blimps, lightplanes, or helicopters. The last has been used for some time by James Bowman of NASA Langley, with motion data radioed to the ground and motion-picture coverage. The time scale for these approximately one-seventh-scale models is very short; but even so, entry and recovery techniques can be investigated. Free-flight-model test results have been found quite generally to agree qualitatively with full-scale results despite Reynolds number differences. But from cost and time considerations, most of these tests have followed, rather than preceded, full-scale tests at high angle of attack.
In design development, drop model testing can be coupled with recently developed parameter identification techniques to supply both the aerodynamic description and an indication of the real airplane behavior at high angles of attack. A large scale model can allow both a better match of Reynolds number and the internal volume required for extensive telemetry equipment and sensors. Telemetry is used to gather detailed data for later analysis and possibly also for modeling an active flight control system. The model can enter stall/departure from maneuvers as well as from l-g flight, and the entire motion from onset through recovery can be experienced. The quickened time scale, however, will not permit a direct pilot evaluation of flying qualities. The model can be used to provide information at extreme flight conditions, with no danger to the pilot. Variations of this technique are currently being pursued by the NASA Flight Research Center (F-15), Air Force Flight Dynamics Laboratory (Lightweight Fighters), NASA Langley Research Center (various models), and the Royal Aircraft Establishment, Bedford, England. At Langley, smaller models with simpler instrumentation and controlled with hobby equipment are being investigated to see how much can be learned from a minimum-cost drop model program aimed at general aviation.
In an attempt to understand the aerodynamics of departure and post-stall motion, flight data are being regressed to determine coefficients of the equations of motion. Pertinent parameter identification techniques range from analog matching of time histories to Kalman filter and maximum likelihood estimates. A conference was held in May 1973 at Edwards AFB, California, to discuss developments in this expanding field. The hope is that this inverse approach will give insight on the form the equations of motion should take. Low-angle-of-attack flight (approximately linear aerodynamics) can now be treated quite satisfactorily, but the nonlinear regions of stall/spin are just now being explored. By applying these techniques to both model and full-scale flight test results, analyses and simulations may be given an updated aerodynamic description of the test configuration. This updated simulation can then be used to plan the remainder of the flight test program more effectively.
If aerodynamic data can be made adequate for the flight conditions to be studied, much can be done to aid design with both linear and nonlinear analysis of the equations of motion. Primary interest is in (1) development of design guides that will help locate potential problems early in the design stage of future aircraft; (2) establishment of quantitative handling qualities criteria for high-angle-of-attack flight regimes; (3) definition of control law for improvement of high-angle-of-attack characteristics through the automatic flight control system; and (4) further identification of good and bad aerodynamic configuration factors and the ways in which they affect the flow and aerodynamic forces.
The increased attention now given to stall/spin characteristics is in itself a major step forward. Relatively simple fixes have been found which, while not eliminating stall/spin problems, have provided significant improvement. Such a fix is the F-111 Stall Inhibitor System, which has been flown. While the flight-test program stopped short of extreme maneuvers, the combination of additional directional stability augmentation and a high stick force gradient in pitch near the limit angle of attack appears to be an effective departure deterrent. For one thing, the stability augmentation forces the airplane to roll more about its flight path than about its body X axis. Leading-edge slats delay both buffet onset and departure of the F-4 to a higher angle of attack. Air Force Flight Dynamics Laboratory's Tactical Weapon Delivery (TWeaD) program demonstrated the effectiveness of improved stability augmentation at angles of attack below departure for a standard F-4. On a simulator, Calspan Corporation has demonstrated that a somewhat more sophisticated change in the flight control system can essentially eliminate A-7D departures; and Vought independently has derived an aerodynamic modification, a wing leading-edge extension next to the fuselage, that gives significant improvement. The F-14 also uses the flight control system to improve departure resistance: "The magnitude of the yawing moment required for a spin cannot be developed when the stick is laterally centered at spin angles of attack."17 A highly successful example of effective control-limiting is the T-38/F-5, which has a potential unrecoverable flat spin mode. While stabilizer authority is adequate for other uses, it is limited to the extent that, in flight test, only abrupt full aft stick held for a long time would develop a spin. Spins have not been encountered at all in T-38/F-5A,B operational use.
We feel strongly that, despite the possible capabilities of a flight control system, there is no substitute for careful aerodynamic design. Nevertheless, other factors may dictate use of the flight control systems to prevent departure from controlled flight. Air combat effectiveness, excessive altitude loss, possible pilot disorientation, and perhaps an unnatural recovery technique are some considerations in further limiting airplane motions.
In summary, then, it can be said that, to improve aerodynamic design capability, the Air Force and NASA are starting to develop better tools. Nevertheless, advancing high-α aerodynamic theory is a long-term project, although intermediate results will be helpful. Studies under way or planned entail fundamental flow theory including viscid-inviscid interaction, with highly instrumented wind-tunnel models to provide data and validation. Also, new dynamic model mounts are in prospect for wind-tunnel testing, and free-flight models are being improved at both ends of the cost spectrum. Simple, less expensive free-flight testing of relatively small models can be helpful early in design, while the effects of stability augmentation, etc., can be investigated with larger, more elaborate free-flight models in time to aid the full-scale flight tests.
Although flying qualities in the normal flight regime can be specified quantitatively in great detail, the complications of nonlinearities and coupled motions near the stall region have precluded the statement of criteria that could be very helpful to the designer. Qualitative statements of general aircraft behavior at the stall region are of limited help. Here again both NASA and the Air Force are starting new research. Ground-based simulators with enhanced visual or motion capability, or both, offer a safe, efficient way to conduct pilot evaluations. Pilot-vehicle analysis has been instrumental in recasting dynamic requirements for lower angles of attack, and now we are attempting to extend these techniques to the stall/spin region. The goal of the flying-qualities efforts is to derive and quantify motion and vehicle parameters and to develop analysis techniques that can be related directly to airplane design.
On the other hand, we realize that "design trade-off" is a synonym of "compromise." Thus we will never be rid of stall/spin problems even if we should come to understand the phenomena thoroughly. But there are several approaches that need to be followed in order to make these design trade-offs possible on a basis approaching rationality.
Air Force Flight Dynamics Laboratory
Notes
1. Orville Wright, Stability of Aeroplanes, Journal of the Franklin Institute, September 1914. Reprinted by the Smithsonian Institution, 1915.
2. A. Glauert, The Investigation of the Spin of an Aeroplane, ARC R&M, no, 618, June 1919.
3. H. A. Sutton, Tail Spins, paper presented at the 18th National Meeting of the SAE, August 1930.
4. Analysis of Aircraft Accidents, Air Corps Information Circular, vol. 4, no. 340, 1 May 1922.
5. Analysis of Aircraft Accidents, Air Corps Information Circular, vol. 7, no. 633, 24 November 1928.
6. Statistical Studies of Aircraft Accidents and Forced Landings, Air Corps Information Circular, vol. 7, no. 652, 28 June 1930.
7. F. E. Weick, The Present Status of Research on Airplane Spinning, paper presented at the 18th National Meeting of the SAE, August 1930.
8. Sutton, op. cit.
9. Wright Air Development Center Airplane Spin Symposium, February 1957.
10. J. H. Wykes, An Analytical Study of the Dynamics of Spinning Aircraft, WADC TR-58-381, February 1960.
11. AFFDL/ASD Stall/Post-stall/Spin Symposium, December 1971 (hereafter cited as AFFDL/ASD Symposium).
12. MIL-S-83691A, Military Specification, Stall/Post-stall/Spin Flight Test Demonstration Requirement for Airplanes, 15 April 1972.
13. MIL-F-8785B (ASG) Amendment 2, Military Specification, Flying Qualities of Piloted Airplanes. (Amendment 2 to be published shortly.)
14. Ibid.
15. M. Tobak and L. G. Schiff, A Nonlinear Aerodynamic Moment Formulation and Its Implications for Dynamic Stability Testing, AIAA paper 71-275, March 1971.
16. W. M. Adams, Analytical Predictions of Airplane Equilibrium Spin Characteristics, NASA TN D-6926, November 1972.
17. AFFDL/ASD Symposium, p. H5.
Other references
First Annual Report of the National Advisory Committee for Aeronautics. U.S. Government Printing Office, Washington, D.C., 1916.
Hunsaker, J. C. Dynamic Stability of Aeroplanes. Smithsonian Institution, Washington, D.C., 1916.
Problems of Aeroplane Improvement. U.S. Naval Consulting Board, August 1918.
Robert J. Woodcock (M.S., Ohio State University) is Principal Scientist, Control Criteria Branch, Flight Control Division, Air Force Flight Dynamics Laboratory, Wright-Patterson AFB, Ohio, where he has performed and directed research and development in flight mechanics, stability and control, and handling qualities of aircraft. He started an advanced development program on stall/spin problems and served until mid-1973 as its first program manager.
Thomas J. Cord (B.S., University of Cincinnati) is an aerospace engineer with the Air Force Flight Dynamics Laboratory. Since graduating in 1970, he has worked in the stall/spin area on programs dealing with high-angle-of-attack criteria and control laws for improved performance at high angles of attack. He currently is project engineer on the Lightweight Fighter Drop Model Program.
Disclaimer
The conclusions and opinions expressed in this document are those of the author cultivated in the freedom of expression, academic environment of Air University. They do not reflect the official position of the U.S. Government, Department of Defense, the United States Air Force or the Air University.
Air & Space Power Home Page | Feedback? Email the Editor