8 CHAPTER 2. MODIFICATIONS TO AIRCRAFT ALREADY CERTIFICATED
5. RE-EVALUATION. Considerable judgment is often required to determine the degree of re-evaluation necessary. If the mass, mass distribution, or the stiffness distribution are affected sufficiently to result in possible significant changes in resonant frequencies of major modes, mode shapes, or mass coupling terms in the flutter equations, then some re-evaluation, such as pre-mod and post-mod Ground Vibration Test (GVT) data comparison, or analysis may be required. Some examples of significant changes are:
a. Engine (Propeller). A change in mass or mass moment of inertia of the powerplant or in its mounting system (bushings, etc.) or a Center of Gravity (cg.) shift should be investigated.
On single-engine airplanes, such changes will most likely affect fuselage and empennage frequencies and mode shapes. For engines mounted on the wings, the entire airplane may be affected.
For changes in existing designs which entail significant increases in engine power, special assessments of the effect on primary and secondary control systems should be made.
If,
(1) Tabs are exposed to the propeller slip stream, And;
(2) the certification basis of the change is to amendment level 23 or later, or;
(3) the changed product rule is applied in order to bring the original certification basis for the change up to amendment level 23 or later, then,
the fail-safe criterion described and discussed in Chapter 3 applies.
b. Structural Cutouts. Severing or bridging across major structural members, such as fuselage bulkheads and ribs or stringers of aerodynamic surfaces, may produce discontinuities in stiffness parameters that significantly alter the vibratory response of the structure.
The significance of a change may be ascertained by flutter analysis.
c. Change in the Mass. Adding of concentrated masses such as a tip tank or an external pod to a lifting surface could have a profound influence on the flutter speeds. This influence needs to be investigated by analysis and/or flight testing.
9 CHAPTER 3. CONTROL SURFACES AND TABS
6. RESPONSE. The aerodynamic force on an airfoil is very sensitive to control surface displacement, which in turn is responsive to both control motions and aerodynamic forces from tab displacement. Control surface displacement may result from deflection of the control system, deflection of the control surface attachment, or structural deflection of the control surface itself under forces from control application, aerodynamic force due to position or rate of position change, and inertia force.
7. BALANCE. Control surfaces and tabs are mass balanced to prevent rotation about their hinges resulting from inertial response to motion of the main (primary) surface, in any flutter mode. When the flutter mode consists of motion about some axis perpendicular to the control surface hinge axis, a concentrated ballast is most efficiently used. Caution should be used to assure that its location is in a high response area of the mode, which is difficult to determine when the mode is complex. Caution should also be used to assure that its attachment is secure to resist the inertia loads. Because the attachment is subjected to oscillatory loads, which cause fatigue failures and because a distributed ballast achieves balance against all flutter modes, it is conservative to distribute the ballast in accordance with the spanwise weight distribution of the surfaces. If less than static balance is provided, the effect of variations in the amount of balance should be evaluated. To guard against unintended balance changes in service, sealing and proper drain holes should be provided to minimize the risk of water, ice, or dirt accumulation in a control surface or tab. Excessive accumulation of these substances could alter the static and/or dynamic balance of the control sufficiently to adversely affect flutter characteristics.
8. VIBRATORY MODES. Control surface rotation about its hinge line is affected by various constraints. Control system stiffness, moment of inertia about the hinge line, and the rigidity of interconnection between control surfaces determine the control surface rotational modes. Both symmetric and anti-symmetric modes should be considered in the vibration analysis. Vibration mode changes resulting from modifications to the control system such as the addition of a bob weight must be assessed for their effect on flutter. Secondary rotations may result from flexure of the attaching structure or bending of the control surface. This is a major consideration for long short-chord tabs and may affect their effective irreversible characteristics. When it is necessary to raise a tab frequency by redesign, consideration should be given to the contributions of: hinge bending perpendicular to the surface especially near the horn-actuator station, horn length, axial stiffness of the push-pull rod or link, mounting flexibility, and lateral stability at push-rod attachment of the tab actuating mechanism.
9. ANALYSES. In most cases involving control surfaces, the flutter speeds are largely governed by the mass balance weights and their distributions. It is wise for the flutter analyst to cover a range of balance values and distributions to determine the most satisfactory ones. It is common to find that a change, which improves one mode degrades another. When conducting a multi-degree-of-freedom analysis, it is advisable to investigate the effect of control system frequency from zero to about 1-1/2 times the system frequency measured in test.
Due to friction, etc., it may be difficult to excite and measure control system frequency accurately. The stiffness can be measured at the surface with the control column or rudder pedal locked in the cockpit and, using the inertia of the end items, the system frequency can be calculated.
10 Theoretical values of tab and control surface aerodynamic derivatives have, for some configurations, produced higher flutter speeds than flutter model testing. Analytically derived tab and control surface aerodynamic coefficients based on strip theory have for some configurations produced higher flutter speeds than wind tunnel tests. Therefore, flutter speed sensitivity to variations in the theoretical coefficients should be evaluated in all control surface/tab investigations.
10. FAIL SAFE REQUIREMENTS. Amendment 23-23 of § 23.629(f) requires flutter free operation after failure, malfunction, or disconnect of any single tab element. This fail-safe requirement is extended to include a failure, malfunction, or disconnect of any element in the primary flight control system or flutter damper on airplanes that do not meet the criteria specified in §§ 23.629(d)(1) through (d)(3).
a. Potential failures that re quire investigation include, but are not limited to, tab or primary control trim actuating system, primary control actuating system (both of which includes bellcranks, pulleys, brackets, and their attachments), and control cables or push rods.
b. Possible means of compliance to actuating sy stem failures (i.e., actuators, cables, rods) may be achieved by incorporating dual systems, mass balancing the controls to counter the rotation of a disconnected or free surface, or by incorporating a combination of the above two.
Proper mass balancing, particularly for tabs, requires considerable care and knowledge of the flutter mechanism to assure adequacy of the design in suppressing flutter. Dual load path designs should include an assessment of residual strength with a single failure to assure that the remaining path will not fail before the single failure is detected during appropriate specified inspection intervals.
c. The question has arisen whethe r the hinges attaching to the control surfaces of the primary flight controls and tabs are part of the control system. Further, are the hinge fasteners and local portions of the fixed-surface structure to which the hinge attaches considered to be part of the control system and therefore to be considered in the failure, malfunction or disconnection of any single element? It could be argued that from a literal or traditional definition, the control system ends at the control horn. On the other hand, if a single element failure of a control surface or tab hinge can cause the surface to flutter, then it may be argued that the objective of the § 23.629(f) rule has not been met.
(1) These questions were investigated thoroughl y in the early 1980s, shortly after the rule change (amendment 23-23) was adopted. The following is a summary of that investigation and the resulting current policy on the matter.
• Interpretation of § 23.651 indicated that the hinges are considered part of the “control surface” rather than the “system”.
• The number of other rules pertaining to hinges and fitting factors that must be applied, generally dictate robust hinge designs.
• A review of service difficulty reports (SDRs) accomplished at that time did not indicate a significant number of failures of control surface or tab hinges.
• Generally, the class of airplanes considered under § 23.629(f)(2), have three or more control surface hinges.
• Tab hinges are often the piano hinge type, which are inherently redundant.
• A review of the regulatory history including the preamble, NPRM, and the 1974- 1975 Airworthiness Review discussions, does not show anything addressing whether the hinges should be included as part of the flight control or tab control system in evaluating § 23.629(f). The implication is that it was not the intent of the rule to include the hinges and supporting structure.
(2) It is, therefore, concluded that the control surface hinges and tab hinges, their attachments, and local portions of structure should not be included as part of the flight control system or tab control system in determining compliance with the flutter requirements of § 23.629(f).
The control surface hinges, tab hinges, their attachments, and local portions of structure are not considered part of the control “system”, regardless of whether the system is fail-safe or not.
(3) In conjunction with this policy, special emphasis should be given to ensure that the rules applying specifically to hinges (§§ 23.393, 23.415, 23.623, 23.625, 23.657, 23.659) are fully evaluated and documented. In addition, hinges should also be evaluated for good design practices with respect to number of hinges, redundancy of fasteners, adequacy of structure adjacent to hinges, methods of providing hinge pin retention if the nut backs off, etc. Similarly, control surface or tab balance weights, their attachments, and local portions of the supporting structure are part of the control surface. They should not be included as part of the primary flight control system or tab control system in determining compliance with the flutter requirements of § 23.629(f).
12 CHAPTER 4. DIVERGENCE AND CONTROL REVERSAL
11. GENERAL. Static aeroelastic instabilities of lifting surfaces are avoided by providing adequate torsional rigidity. Methods to determine the adequacy of torsional rigidity are outlined in references 2 and 3 of Appendix 3.
12. DIVERGENCE. Divergence occurs when the aerodynamic torque exceeds the torque resisting capability of the wing. Because the aerodynamic torque is a function of speed as well as deflection, whereas the resisting torque depends on the torsional rigidity of the lifting surface which is a constant, there exists a limiting divergence speed. Divergence may occur with no warning.
13. CONTROL REVERSAL. Control reversal will often be preceded by pilot comments of “heavy” or “sluggish” ailerons. A limiting reversal speed is reached when the change in lift due to control surface rotation is nullified by the change in lift due to twist of the lifting surface.
A1-1 APPENDIX 1. GROUND TESTING
1. INTRODUCTION. The adequacy of the methods used to show compliance with § 23.629, as discussed in the main body of this document, is dependent upon the availability of reliable ground test data to verify the analytical data used and/or to serve as a basis for flutter substantiation per the simplified criteria of reference 1. This appendix, therefore, presents guidelines in conducting the more significant tests required to accomplish this objective. However, in keeping with the general purpose of this advisory circular, the information provided is not intended to be mandatory, nor is it to be considered an exhaustive treatment of the subject.
2. CONTROL SURFACE AND TAB MASS PROPERTIES. The experimental mass properties of control surfaces and tabs (weight, static moments, moments of inertia, and
c.g.) are important ingredients in flutter substantiation. These properties form a basis for verification of the analytical data used in the rational analysis and provide the necessary parameters for use in the simplified criteria. Reference 1 presents a detailed procedure for the experimental determination of these properties.
3. TAB FREE PLAY. Free play tests provide the necessary data for determining the effectiveness of a tab in fulfilling the requirements for irreversibility as specified in the main body of this document. In addition to demonstrating the maximum free play available, these tests provide the stiffness of the actuating system for use in computing tab rotational frequency.
a. Free play and stiffness may best be measured by a simple static test wherein “upward” and “downward” (or “leftward” and ”rightward”) point forces are applied near the trailing edge of the tab at the spanwise attachment of the actuator (so as not to twist the tab). The control surface should be blocked to its main surface. Rotational deflection readings are then taken near the tab trailing edge using an appropriate measuring device, such as a dial gauge. Several stepwise load and deflection readings should be taken using loads first applied in one direction, then in the opposite.
b. A plot of these load deflections typically appears as shown in Figure A-1.
Figure A-1. Load Deflections Plot
A1-2
c. Free play is then defined by extending th e best straight lines through zero. System stiffness may then be obtained from the slopes of the curves away from the zero point.
4. INFLUENCE COEFFICIENT TESTS. Bending and/or torsion influence coefficient test results form the basis for the definition of component stiffness distributions. The extent of the tests depends on the intended use of the data. A full scale test program, wherein the coefficients of each spanwise mass strip are defined, may be desired if experimental data is the primary source for defining component stiffness. In contrast, calculated influence coefficients, based on analytical bending (EI) and torsion (GJ) stiffness distributions, may be adjusted reliably with considerably less test data. A method is outlined below for determining influence coefficients for conventional structure, i.e., aspect ratio greater than four and unswept elastic axis.
a. The test article, wing, tailplane, or fin, is generally mounted at its root, without control surfaces, in a rigid test fixture for these tests. However, wing stiffness tests, particularly torsion as required for simplified criteria, may be successfully conducted with the wing mounted on the fuselage restrained in a cradle. This type of setup requires duplicate loading fixtures for right and left wing to balance the aircraft under load and thus minimize “jig rotation” effects.
b. The chordwise location of the elastic axis is determined by applying a torque load at selected stations and plotting the deflection versus chord shear center or elastic axis at that station.
c. Torsional influence coefficients (radians twist about the elastic axis per unit torque load) are obtained by applying a pure torque load about the elastic axis at the tip and measuring the resulting spanwise twist. The twist per unit torque applied at intermediate inboard stations will be the same inboard of the load point. Thus, it is necessary to load only one additional inboard station, say 75 percent span, to check for data repeatability only. To insure that the load applied is a pure torque load, the deflections of the elastic axis should be monitored during the loading process. Zero deflections should result.
d. Bending influence coefficients (deflecti ons per unit shear load) are obtained by applying shear load on the elastic axis at a selected station and measuring the resulting deflections at a sufficient number of spanwise locations to define the influence line for that load point. The procedure is repeated for each load station. To insure that the shear load is applied on the elastic axis, no appreciable chordwise variation in the measured deflections should be evident.
e. The experimental determination of fuselage stiffness properties can be accomplished essentially the same way as for the aerodynamic surfaces. In this test the fuselage is treated as two beams, forward and aft fuselage, each cantilevered from the wing-root attachment. It is extremely important that the fixture at this attachment be very rigid; and, any displacement of the test jig during loading must be monitored, regardless of how small, throughout the test for inclusion in the data analysis. Small displacements can be quite influential in a rather complex data reduction procedure, and if improperly done, can lead to erroneous and troublesome conclusions. On this basis it is often the practice to compute fuselage stiffness properties for the fuselage, then use ground vibration test results
A1-3 to tune calculated modes and, in turn, stiffness as required.
f. Thin-skinned structure may buckle at a ve ry low load, reducing actual stiffness in flight considerably from that determined by the above procedure and the analyst is cautioned to investigate such conditions.
5. GROUND VIBRATION TESTS. Ground vibration testing has as its fundamental objective the definition of vibration mode frequencies, mode shapes, and damping characteristics of an aircraft. These data then become the basis for the analytical development of a mathematical vibration model of the airplane or serve as a check on such a model once it is developed. The results ultimately become the basis for rational flutter analyses. If the simplified flutter prevention criteria of reference 1, discussed in the main body of this advisory circular, is used, then the results from these tests are used directly to establish a predicted flutter speed of the airplane.
The degree of sophistication required to conduct a resonance test (techniques, recording equipment, suspension system, etc.) depends upon the complexity of the structure being tested. Since it is impossible to cover all test situations that may arise, the discussions presented in this section are fundamental in nature, dealing specifically with sinusodial methods of excitations. They are intended as guidelines for those persons concerned with general type aircraft, which have only the basic test facilities. Other procedures employing random or impulse excitations are being used more frequently. However, these methods are considered beyond the scope of this advisory circular.
a. Test Article and Suspension System. The airplane should be supported in a level attitude such that the rigid body frequencies of the airplane on its support are less than onehalf the frequencies of the lowest elastic wing or fuselage mode to be excited.
One of the following methods of support can generally be used:
(1) Support the airplane on its landing gear with the tires deflated sufficiently to achieve the above result. Fifty percent normal tire pressure usually achieves good results.
It may be necessary to block the landing gear struts to eliminate damping in the oleos.
(2) Suspend the airplane on springs.
(3) Support the airplane on its landing gear resting on spring platforms.
(4) Support the airplane fuselage and wi ngs on large air-filled flotation bags.
(a) The airplane should be equipped wi th all items having appreciable mass such as engines and tip tanks. The weight and c.g. of the test article should be determined to enable proper correlation with the analytical model. Where fuel is located in the outboard 50 percent of the wing semispan, it may be desirable to test a full fuel condition in addition to the empty condition in order to provide additional data for math model correlation.
(b) It is generally advantageous to lock the control surfaces in their neutral position when obtaining airframe modes.
A1-4
b. Equipment. Various types of shakers are available, i.e., inertia, elastic, airjet, electromagnetic, etc. Electromagnetic exciters are generally preferred and most commonly used. This type consists of a coil that is attached to the structure with a fixed drive rod, as opposed to a flexible shaft or spring for inertia or elastic type shakers. The coil is surrounded by a magnetic field and is set in motion by an alternating current. Electronic oscillators and amplifiers are used to control this type of system.
(1) Vibration amplitude may be obtained by using either velocity pickups or accelerometers so long as transducer mass is insignificant. Most modern accelerometers weigh only a few ounces. Typically that is insignificant when installed on the fuselage, wings, or vertical and horizontal stabilizers. When used on control surfaces, this same accelerometer may affect the flutter characteristics. Care should be exercised to select a small, light-weight accelerometer when mounting on control surfaces like tabs. The output can be observed using a cathode ray oscilloscope and digital voltmeter. Phase relationship between two transducers can be noted with sufficient accuracy, and by exercising extra care, using an oscilloscope equipped with a grid screen.
(2) Data systems are available that provide the coincident, in- phase or real term, and the quadrature, the imaginary term, responses of the total response frequency (the product of the force and reference signal). Graphical representation of these terms is presented, providing a very accurate identification technique for resonant frequencies and phase relationships. Structural damping is also readily available from these data.
(3) Whatever data system is use d, uniformity is recommended. Piecemeal systems, using velocity pickups and accelerometers, or filters with different characteristics, etc., can give erroneous data and should not be used without careful regard to their calibrations and performance characteristics and limitations.
c. General Procedures for Airframe Modes. It is usually sufficient to apply a harmonic excitation force to the structure provided the force is not applied in the proximity of a node line. For this reason vibrators are usually attached at an extremity such as the nose and/or rear of the fuselage or near the tips of the wing or empennage surfaces where nodes are not likely to occur.
(1) The node line lies along th e elastic axis of the structure (fuselage, wing, stabilizer, etc.). The axis passes through the loci of the shear centers of the cross-sections along the span of the structure. This theoretical location is typically determined mathematically when the dynamic computer model is created, and may be tweaked during the ground vibration tests. It can also be determined during the vibration tests, since an excitation force that is applied in the proximity of a node line will produce a near zero structural response. The point of application of the excitation force can then be moved away from that point and tried again.
(2) With the shaker(s) and a reference pickup mounted at a selected location, frequency is varied upward through the range usually encountered in aircraft structures.
With small increments of frequency, the response of the structure is recorded and the resulting plot of amplitude of response versus forcing frequency is used to determine the resonant frequencies of the system. A typical sweep is shown in Figure A-2.
A1-5
Figure A-2. Typical Sweep
(3) Although duplication of peak responses will result, it is advantageous to obtain frequency response records with a reference pickup positioned on each of the main surfaces and fuselage at a specific shaker location. This will reduce the chances of overlooking modes.
(4) There are several criteria for establishing that the excited response approximates a normal mode of vibration. The most commonly accepted approach requires that all of the criteria below be met:
(a) A relative maximum response per unit input exists. A normal mode of vibration occurs at a resonant frequency. At resonant frequency, the response is at its maximum. One needs to “tune” the excitation frequency very carefully until the maximum response is reached. The frequency at which the maximum response is observed, gives the resonant frequency. If the excitation frequency is slightly off of that frequency, the response will fall well below the maximum response.
(b) Accelerations at all points in the structure are either exactly in phase or 180 degrees out of phase with each other. The accelerations measured at all points on the structure during resonance will be either in phase or out of phase with a reference location but will be at a ± 90 degree phase angle with the force, for small values of damping.
(c) A decay record exhibits a single-frequency, non-beating, low-damped characteristic.
(5) Having established the resonant frequencies, a survey of the aircraft is conducted with the shakers tuned to each frequency in-turn. A roving transducer is used to sense amplitude and phase angle relative to the reference pickup at each airplane location.
An adequate number of points should be surveyed along the span and chord (typically on the spars) of each surface and along the fuselage to define the airplane modal displacements, and the associated node lines. The number of survey points depends on the nature and span of the surface and the number of modes to be surveyed. If one is interested in obtaining a number of modes, (like third bending or fourth bending modes) more accelerometers are needed span-wise. Usually 8 to 10 accelerometers are sufficient for a small airplane. To obtain proper phase relationship additional excitation may be necessary.
A1-6
(6) It may not be necessary to survey identical peak frequency responses although they occur at different locations. In all probability, the mode will be the same. This can be determined by checking only a few stations or simply by visually observing the motion of the aircraft.
(7) Care should be exercised in defi ning component node lines for each mode.
This is particularly important in evaluating the effectiveness of balance weight locations.
d. Aircraft Structural Modes Usually Encountered. The modes excited during ground vibration depend on the type of configuration being tested. The vibration modes of an airplane that carries heavy mass on the wing, such as engines, tip tanks, etc., or has the stabilizer located high on the fin will be highly coupled and generally cannot be described except by diagrams that show the relative shape and phase of each part of the airplane.
Airplanes that do not have these design characteristics usually have relatively uncoupled modes, which can be described by naming the type of motion that is predominant. In general, the following predominant modes should be obtained insofar as is practicable.
(1) Wing Group Modes.
(a) For wings without engines, tip ta nks, or heavy external or internal stores: Wing vertical bending and wing torsion, fundamental and higher modes, symmetric and anti-symmetric.
(b) For wings carrying heavy masses outboard of the fuselage: Wing bending coupled with wing torsion and flexible store (engines) modes, fundamental and higher modes, symmetric and anti-symmetric.
(2) Fuselage - Empennage Group Nodes.
(a) Fuselage Torsion (coupled with stabilizer anti-symmetric bending).
(b) Fuselage lateral bending and fi n bending, fundamental and higher order consisting of two fundamental modes in which the fin tip and aft fuselage are in phase in one mode, and out of phase in the other.
(c) Fin bending - symmetric and anti-symmetric for multi-tail airplanes.
(c) Fin torsion (generally highly coupled with stabilizer yawing if stabilizer is located at the outer span stations of the fin).
(e) Rudder bending and torsion.
(f) Fuselage vertical bending and stabilizer bending, fundamental and higher order consisting of two fundamental modes in which the aft fuselage and stabilizer tips are in phase in one mode, and out of phase in the other.
A1-7
(g) Stabilizer torsion - symmetric and anti-symmetric.
(h) Stabilizer yawing for surface located at the outer span stations of the fin.
(i) All movable horizontal tail - ro tation coupled with bending, torsion.
(3) Engine or External Store Modes. For multiengine aircraft or aircraft carrying “large” pylon-mounted stores (the weight and the cg location with respect to the elastic axis of lifting surface are the significant factors in this instance), the pitch, roll, yaw, and lateral and vertical translation modes should be defined. These modes should also be determined for all turbo propeller engine installations. It may be necessary to excite the engine fore and aft on the propeller blade to obtain the most critical pitch and yaw modes. If this method is used, consideration should be given to possible modal distortion due to propeller blade flexibility. Also, caution should be exercised and the engine manufacturer's instructions followed concerning possible damage to bearings when exciting the engine.
e. General Procedures for Control System Modes. The experimental determination of control surface and tab rotation modes about their hinge lines may be difficult due to inherent friction within the system or the masking of these modes by structural interaction.
(1) On this basis, extra care is required for proper identification of the system's characteristics.
(2) For conventional aileron or elevator systems, the rotation modes may be successfully measured by applying a single excitation force to either the right hand or left hand surface. However, multiple shakers are preferred, particularly if the right and left surfaces are operated from separate control systems. Likely shaker positions are on the trailing edge at midspan or on the horn leading edge. Tab rotation may be determined from the control surface excitation but usually a direct excitation on the tab surface is required with the control surface (aileron, elevator, or rudder) blocked to its main surface.
(3) A transducer placed on the control being excited is used to monitor the response and determine peak frequencies by the same technique described for airframe modes. To define the modes excited, it is generally necessary to follow any or all of the following procedures:
(a) Monitor the phase between the right and left surface, the control column, or the attaching structure.
(b) Conduct a detailed survey of the surface, spanwise and chordwise, to define any structural modes. If the surface has a very long span or wide chord, these modes, bending and torsion, are likely to be dominant.
(c) Visually monitor the surface under excitation.
(d) Simple rationalization to distinguis h the excited modes from previously defined airframe modes.