A2-1 APPENDIX 2. FLIGHT FLUTTER TESTING
1. INTRODUCTION. This appendix presents a general discussion of acceptable procedures for conducting flight flutter tests intended as final validation of flutter free operation within the flight envelope for new or modified airplanes. The methods described herein do not represent a comprehensive survey of existing techniques, but rather represent methods, which have been proven to be particularly adaptable to general aviation aircraft.
a. It is recommended that these tests be conducted only after appropriate analyses, defining the critical conditions and severity of flutter onset, have been performed. Both the risk and scope of testing required to substantiate the total airplane is significantly increased without the benefit of reference analyses.
b. In-flight excitation of only the critical mode (s) is generally, all that is necessary for final demonstration of flutter free operation if preceded by rational flutter analyses.
However, without these studies, all modes of the airplane, or those modes affected by the modification of an altered airplane, must be excited in flight. All test airplanes should include proper instrumentation for recording airplane response.
2. DETERMINATION OF VIBRATION CHARACTERISTICS. Section 23.629(a) requires a determination of the natural frequencies of main structural components by vibration tests or other approved methods. This must be done regardless of the flutter substantiation method selected; i.e., (a) rational analysis plus flight flutter tests, or (b) simplified criteria plus flight flutter tests. This determination must be made for all new airplanes and for existing airplanes, before and after any major modification to assess the effects of these structural changes. Engineering judgment should be exercised in determining whether the effects of the modification on aerodynamics, stiffness, or mass are sufficient to warrant flutter reinvestigation. The effects of variations in fuel loading, airplane weight, and center of gravity should also be assessed. It is recommended that mode shapes, as well as frequency, be determined by either ground vibration tests or analytical methods, if adequately supported by test.
For modified airplanes with no available analyses, the degree of frequency change requiring flight investigation is dependent upon, in part, the nature of the modification, the relative change of bending-torsion frequency ratios, and the relationship of the structural modes to control surface modes. Shifts in node lines may also dictate flight checks.
3. AIRCRAFT EXCITATION METHODS. The airframe modes and frequencies can be excited in flight by any number of techniques. The important criteria for technique selection is that the modes and frequencies of interest must be adequately excited to allow for proper modal response. During the tests, flutter critical modes are to be excited. If an 11-Hz. mode is critical, the method of excitation must be positively able to excite that mode. The flight test data should be able show that mode at that frequency and it should be distinct in the frequency-amplitude plot. Methods like stick raps will not excite frequencies higher than 7 to 8 Hz. In such cases, eccentric mass excitation or other methods must be used. To illustrate, most general aviation aircraft use cable or push rod control systems,
A2-2 which have high levels of coulomb damping. The coulomb damping will cause a nonlinear control response. At low amplitudes the damping will be high and the system stable;
whereas, at higher amplitudes, the coulomb damping will be reduced and the control system could be unstable. A proper level of modal excitation will, therefore, produce lower system damping and an earlier indication of a developing flutter mechanism.
Consequently, without proper excitation, the test engineer may have very little warning of developing flutter.
Consideration should also be given to the possible influence of the excitation system on the flutter characteristics of the airplane. Excitation methods presently being used for general aviation airplanes are discussed briefly in the following paragraphs.
a. Pilot Induced Control Surface Impulses. Pulsing the controls may be used to excite modes, generally, below 8 Hz, and is not recommended above 8 Hz. The effectiveness of the method is usually limited by either the ability of the pilot to impart a pulse of proper duration or the ability of the control to transmit the pulse to the primary surface. Three degrees of control rotation are normally sufficient if the duration is short enough to encompass all harmonics of interest.
b. Sinusoidal Excitation Using Rotation Masses. This technique has been used successfully for exciting airplane modes between 10 and 50 Hz. Rotating eccentric mass shakers mounted in the wing tips and/or the tail section of the fuselage will usually produce wing and empennage modes of concern. Although shakers in each wing tip and both vertical and lateral mounted fuselage shakers may be required to excite both symmetric and anti-symmetric modes, single shakers have been used successfully for both symmetries where adequate frequency separation exists. The eccentric mass should be large enough to excite the 10 Hz modes and the shaker supporting structure strong enough to withstand the shaker force at 50 Hz. Shaker forces of up to 300 pounds at 50 Hz have been used and produced very good results. In general, the larger the exciting forces the safer the test since the non-linearity will be minimized. The shaker force should be adequate to assure that modal response is easily distinguishable from random or buffet excitation. In other words, the response should be well above the noise levels of the data. This can only be determined during the test. If the response is not distinct, the shaker force is to be increased suitably.
Several samplings of data will eliminate the random and buffet responses.
A simple inertial shaker system can be constructed using off-the-shelf shop components.
The system consists of a container of compressed nitrogen, pressure regulator, line gate valve, air hose and shop drill motor. The RPM of the drill motor is controlled by varying the line pressure and can be monitored by a tach generator attached to the drill motor shaft.
The only parts requiring design and fabrication are the eccentric mass and appropriate mounting brackets.
c. Excitation Using Autopilot And Other Methods. Sinusoidal excitation using the autopilot will produce modal responses similar to the inertial system but has the advantage of supplying a stronger input to the fundamental modes. A restriction is its ability to transmit energy into the control surfaces at the higher frequencies due to control system flexibility.
A2-3 Other methods such as flutter vanes and rocket impulse units may be used. Regardless of the method used, the same principles of frequency response and mode identification apply;
i.e., adequate response of the modes and frequencies will produce the desired indication of any developing flutter mechanism.
4. AIRCRAFT INSTRUMENTATION. The aircraft instrumentation required to adequately monitor vibration characteristics will vary greatly depending on the extent of the test (number of modes being investigated) and the special design characteristics. As a minimum, transducers that measure acceleration or velocity should be installed on the tips of the aerodynamic surfaces of concern; i.e., for wing and horizontal stabilizers, front and rear spars on one side for bending and torsion response and on one spar of the opposite side as an indicator of symmetry. The frequency response characteristics of the installed transducer should be checked to assure adequate sensitivity throughout the test frequency range. Strain gages or accelerometers should also be installed on each control surface.
Care should be taken to assure control balance is not disturbed by the instrumentation.
This may be accomplished by choosing light-weight accelerometers, and avoiding placement of the accelerometers at the leading or trailing edge of the control surface.
a. Unless workload makes it prohibitive, it is preferred that the in-flight data recorder and exciters be operable by the pilot to eliminate any need for additional personnel on board the airplane during the test. This is particularly desirable if testing is performed without supporting analysis.
b. Although various telemetry systems exist for recording and transmitting data to the ground, an on-board oscillograph or magnetic tape recorder are more commonly used. A magnetic tape recorder is advantageous over an oscillograph if filtering, sensitivity change, and speed variations are available on playback to aid in determining frequency and damping. An oscilloscope, or other means to monitor input frequency in real time, should also be available to the crew.
c. The complete excitation and data reco rding system should be thoroughly checked and calibrated, on the ground and as mounted in the airplane, to assure excitation of the desired modes and in turn to establish baseline amplitude and damping data. The response to engine noise and aerodynamic buffeting can seriously distort the data and, as a general rule, the signal-to-noise ratio should be at least 4 to 1. The unwanted signals can be filtered or minimized by increasing the exciter force. However, the maximum level of excitation should be limited to prevent structural damage from dynamic overload.
5. TEST CONDITIONS.
NOTE
Several of the following are recommendations for safety only (noted as *Safety Recommendation*) and not required for showing compliance. The applicant is free to pursue other means of achieving a safe, flight test program.
a. Flight flutter tests shoul d be conducted with the airplane configured to provide maximum safety to the crew. In preparing the airplane, a checklist of configuration requirements should be closely adhered to. The list should include the following.
A2-4
(1) The airspeed indicating system should be calibrated.
(2) All equipment items in the cabin should be secured adequately to meet emergency landing load requirements of § 23.561. *Safety Recommendation*
(3) When certification is by § 23.629(d) (sim plified criteria) and testing, control surfaces should be balanced to the most under balanced (tail heavy) condition and trim tab free play set at the maximum allowed. If certification is by analysis and testing, the analysis should dictate the appropriate settings.
(4) Control system damping should be minimized to simulate wear.
(5) The crew should be provided with parachutes. *Safety Recommendation*
(6) Each of the crew members should have easy access to an escape exit.
*Safety Recommendation*
(7) Each emergency exit door should be equipped with a quick release mechanism allowing the door to separate from the airplane. The doors should be checked at various airplane yaw angles to make sure that the pressure distribution over the door will allow it to be drawn away from the airplane. *Safety Recommendation*
(8) For airplanes with larg e cabins, a knotted rope should be installed the length of space of the cabin. *Safety Recommendation*
b. The airplane configuration(s) to be tested (i.e., fuel loading, c.g., weight) obviously depends upon the purpose of the test, upon whether the airplane is a new type design or a modified type design, and upon the nature of the modification if the airplane is a modified type design, etc. Generally, a minimum of two-wing fuel loading conditions, representing maximum and minimum, should be checked if this parameter affects a significant change in component modes. Acceptable tolerances from the selected condition, to account for fuel usage, must be established and maintained throughout the tests. The need to explore airplane weight and c.g. variations must be based on the effect of these parameters on the airplane component(s) of concern.
c. At least two altitudes should be ch ecked, if appropriate to the design. One representative of high dynamic pressure (approximately 9,000 feet will provide emergency egress protection) and one at approximately 75-percent service ceiling or the altitude at the Vd/Md knee, if a design limiting condition. These recommendations may not be appropriate or adequate for all airplane designs. As one example, a 9,000-foot altitude would not be safe for a Mach .72/275 kcas airplane that descends at 10,000-feet per minute to reach dive speed. Qualified individuals must make the final determination of the sufficiency of the test procedures to demonstrate compliance in a safe manner.
6. FLIGHT TEST PROCEDURES. Due to the potential dangers involved in conducting flight flutter tests, there is always the desire of the crew to get the tests over with and a general tendency to want to skip speed points and short cut the test. For this reason, a flight plan should be established prior to beginning the test and followed as closely as possible.
A2-5
a. A chase plane should be used for all test s. It is recommended that the chase plane have videotape capabilities. This is a safety recommendation only, and is not required for showing compliance, however it is a standard practice.
b. Testing should begin at a low airspeed poi nt to establish a database; i.e., frequency correlation with ground vibration modes. Follow-on increments should be based on the flight history of the airplane. If the airplane has previously flown to some limit speed, then enough test points should be checked to that pre-achieved speed to develop a data trend.
For unexplored speeds, the increments should be smaller and no more than two-speed data points checked at any one altitude during a flight unless data trends show continuously increasing, or very high, damping with no indications of ensuing shifts in this trend. More data points may be checked per flight if telemetry and real time data analysis systems are available. Between VC and VD, the number of airspeed increments should be increased.
c. Atmospheric turbulence should be avoide d as much as possible during these tests to eliminate superposition of unwanted random signals on the data records and to preclude possible structural overload at speeds near VD.
d. At high speeds, prior to the commencement of test points, the test pilot should establish adequate controllability of the airplane by slowly exercising each of the three control systems and ensuring the proper response is received.
e. If inertia shakers, aero vanes, or other sinusoidal exciters are used, frequency sweeps should be conducted at each speed point. It is also recommended that shaker dwells be performed at selected airspeeds as a check on the damping characteristics established from sweeps. These are conducted by tuning the shaker(s) to each peak frequency, allowing time for the airplane to stabilize, then cutting the shaker power and recording the response of each transducer. If pilot induced control impulses are used, each axis should be pulsed at least twice in each direction wherein the airplane is allowed to stabilize, with hands off controls, prior to the next impulse or speed point.
7. DATA REDUCTION AND INTERPRETATION. The methods used for analyzing flight flutter data will depend on (1) the type of excitation used, (2) the availability of electronic analysis equipment, (3) the degree of accuracy required, and (4) the time allowed for data reduction. Generally, absolute damping values are not necessary to achieve the objective of the flight tests wherein monitoring of damping versus airspeed trends is the primary concern. The approximation methods addressed briefly in this section will allow development of reliable trends provided consistent procedures are followed in reducing the response traces throughout the test.
a. When impulse excitation is used, th e damping can be obtained by measuring the decay rate directly from the response traces. For cases where several frequencies are being excited by the impulse, it may be necessary to reverse the transient and play it into a tuned filter. The frequency of the tuned filter can then be varied to yield discrete modal responses (reference 7).
b. When continuously forced oscillation techniques are used to excite the structure, either the amplitude response method or the vectorial analysis method as developed by Kennedy and Pancu (reference 6) can be used to reduce the data. The amplitude response
A2-6 method is advantageous for general aviation applications since it provides a good approximate damping level, requires a minimum of electronic equipment, and the data can be quickly reduced.
For this method it is assumed that the relative damping ratio is approximately inversely proportional to the maximum resonant amplitude of the respective modes. Therefore, if the amplitude for a given mode is increasing as the airspeeds increase, a reduction in stability will be indicated. The approximation method is briefly outlined as follows: (1) The product of response amplitude times damping will be constant for a given exciter force; i.e., (Ag) (gg) = c
(2) If the exciter force relative to frequency remains constant, then at a given airspeed, the net structural damping will be:
(Av) (gv) = (Ag) (gg)
then:
gv=(Ag) (gg)/Av
where:
A g=peak amplitude on ground A v=peak amplitude in air at test velocity g g=damping measured on ground at zero velocity g v=net structural damping at test velocity
(3) The net damping (g v) can be defined as the actual structural damping measured on the ground (gg) minus the analytical aerodynamic damping (g) determined from the flutter analysis with zero structural damping; i.e., gv=gg-g, where g is negative when stable.
(4) To permit use of this method, the shaker force and sweep rate must remain consistent from run to run.
c. A gradual increase in peak response (A v) by a factor of 3 or a rapid increase in Av by a factor of 2 will normally require stopping the test to inspect the airplane, instrumentation, etc.
For example, a gradual increase in response occurs over a large range of speed (200 to 300 knots). Therefore, an increase from 10 units response at 150 knots, to 30 units response at 350 knots may be considered a gradual increase in peak response. An example of a rapid increase may be 20 units response at 250 knots, to 40 units response at 280 knots. One must rely on engineering judgment to make this determination. It is also possible to experience a very sharp rise in damping followed by a sharp decrease leading to violent flutter, thus making it difficult to predict trends without the aid of reliable analyses.
d. It is advisable to check the effects of variation in sweep frequency on the peak response damping using techniques included in Appendix 3, reference 2. It is also advantageous to supplement continuously forced damping measurements with damping obtained from decay records. Flutter margin predictions per the techniques of Zimmerman and Weissenburger (Journal of Aircraft, 1964, Volume 1, Number 4) may be a beneficial data presentation approach to supplement the characteristic velocity versus damping plots.
A3-1 APPENDIX 3. ACKNOWLEDGMENTS, REFERENCES AND BIBLIOGRAPHY
REFERENCES:
1. Airframe and Equipment Engineering Report No. 45, Simplified Flutter Prevention Criteria for Personal Type Aircraft." FAA, Engineering and Manufacturing Division, Flight Standards Service, Washington, D.C.
2. Introduction to the Study of Aircraft Vibration and Flutter, Scanlan and Rosenbaum, McMillan Company, New York, 1951.
3. Aeroelasticity, Bisplinghoff, Ashley, and Halfman, Addison-Wesley Publishing Co., Reading, Mass., 1955,
4. Application of Three-Dimensional Flutter Theory to Aircraft Structures, Air Force Technical Report (A.F.T.R.) 4798, 1942.
5. Tab Flutter Theory and Application, Air Force Technical Report, A.F.T.R., 5153, 1944.
6. The Use of Vectors in Vibration Measurement and Analysis, Kennedy and Pancu, Journal of Aeronautical Sciences, J.A.S., Vol. 14, 1947.
7. A Survey of Flight Flutter Testing Techniques, AGARD Flight Test Manual, Vol. 11, Chapter 14.
8. Mechanical Vibrations, Thomson, Prentice-Hall, New York, 1953.
9. An Analytical Treatment of Aircraft Propeller Precession Instability, Reed and Bland, NASA TN D-659.
10. Propeller Nacelle Whirl Flutter, Houbolt and Reed, Journal of Aeronautical Sciences (J.A.S.) March, 1962.
11. Review of Propeller-Rotor Whirl Flutter, NASA TR-264, July, 1967.
12. Cantilevered Power Package Whirl Mode Flutter Program, G.S. Rasmussen & Associates, Report No. 650, June 1, 1971.
13. Wind-Tunnel Measurement of Propeller Whirl-Flutter Speeds and Static Stability Derivatives and Comparison with Theory, NASA TN D-1807, August 1963.
A3-2
14. Laguerre's Method Applied to the Matrix Eigenvalue Problem Beresford Parlett, A Paper Prepared Under Contract AT (30-l)-1480 with the U.S.A.E.C.
BIBLIOGRAPHY:
1. Manual on Aeroelasticity in Axial-Flow Turbomachines, North Atlantic Treaty Organization Advisory Group for Aeronautical Research and Development (AGARD), 1987.
2. Analytical Methods in Vibration. Mierovetch, McMillan Company, New York, 1967.
3. An Introduction to the Theory of Aeroelasticity, Fung, John Wiley and Co., New York, 1955.
4. Dynamics of Structures, Hurty and Rubinstein, Prentice-Hall, New York, 1964.
5. Sane Aspects of Ground and Flight Vibration Tests, Mazet, ONERA TN No.
34, 1956.
6. A Survey of Aircraft Subcritical Flight Flutter Testing Methods, Rosenbaum, A.R.A.P. Report No. 218, 1974.
7. Subcritical Flutter Testing and System Identification, Houbolt, A.R.A.P. Report No. 219, 1974.
8. Proceedings of the Flight Flutter Testing Symposium, May 1958, Washington, D.C., O.S.R.- 9-0269.