The clearance process relating to aeroelastic and loads issues during the design and certification •phase of a new aircraft is a combination of numerical modelling backed up by testing.

A similar series of tests are undertaken to certify aircraft for ground and flight loads. •

No one test is capable of providing the information for full validation of the mathematical models •used for certification. That must inevitably come by building up a range of test results on different aspects (e.g. structural stiffness, mass, mass dis- tribution, centre of mass, wind tunnel tests, systems tests, etc.).

The ground tests can be quite accurate and are backed up by checks that show, when assembled •together, that the structural dynamic and the flight response properties are both reasonable.

However, tests performed in- flight to demonstrate aeroelastic stability and validate flight loads are •subject to a number of uncertainties.

The test set up for the Flight Flutter Test is much less ideal (e.g. noisy environment, inadequate •excitation) than that for the Ground Vibration Test, whereas flight loads have to be determined from strain gauge readings and these calibrations can be problematic.

Wind Tunnel Tests

There are two types of wind tunnel test that are particularly relevant to aeroelasticity and dynamic •loads:

determination of rigid aircraft aerodynamic derivatives◦

flutter model testing. ◦

Testing Relevant to Aeroelasticity and Loads

Aircraft configurations are often complex and accurate prediction of rigid aircraft aerodynamic •derivatives is difficult, especially when non-linearity is important (e.g. near to the stall condition) or when the aircraft is in the transonic regime.

Model tests for the rigid scale model of a whole aircraft are performed to measure pressure •distributions, net forces and moments and so estimate the aerodynamic derivatives at different flight conditions.

Values obtained are used to validate and possibly update results calculated using design formulae, •data sheets, CFD, etc., and also to scale calculated unsteady aerodynamic results at zero reduced frequency.

A wind tunnel flutter test program (AMC 25.629) may sometimes be undertaken using a dynamically •scaled model:

However, such testing is not sufficiently reliable to use in clearing the aircraft for flutter◦

Instead it helps in the main certification calculations by validating unsteady aerodynamic ◦methodologies, performing parametric studies, studying new configurations, investigating interference and compressibility effects, etc.

Ground Vibration Test

The Ground Vibration Test or GVT, sometimes called a modal test (Ewins, 1995), is performed on the •prototype aircraft to obtain estimates of the whole aircraft normal modes (natural frequencies, damping ratios, mode shapes and modal masses).

These modal data may then be used to confirm, or adjust, the calculated normal modes used in the •critical flutter calculations, as well as providing substantiated estimates of damping.

Different fuel and hydraulics configurations are studied.•

The aircraft is usually supported on soft springs (e.g. air bags, elastic bungees or deflated tyres) so •that it behaves as near to the free –free condition as possible.

Alternatively, where this is not possible (e.g. where the aircraft has very low frequency modes), a stiff •support may be employed. Whatever the support arrangement, it may be included in the dynamic model when comparing measurements with calculations.

The aircraft is instrumented with, typically, many hundreds of accelerometers to allow adequate •mode shape definition, and a number of electrodynamic exciters (typically up to eight) are used to excite vibration of the aircraft at a sufficiently high level.

Normally, multiple exciters are applied simultaneously to distribute energy adequately over the •structure and to allow modes close in frequency to be identified; exciter positions may need to be varied to excite modes successfully.

The excitation signals used to drive the exciters are usually sinusoidal or random, depending upon •the test methodology being employed.

There are two main approaches to testing:•

Phase separation method: ◦

A broadband multipoint uncorrelated random excitation or swept sine correlated excitation ‣is applied to the structure using several exciters.Then a matrix of Frequency Response Functions (FRFs) is estimated and a frequency or ‣time domain identification (i.e. a curve fit) is employed to identify the modal properties.

Phase resonance method: ◦

Sinusoidal excitation forces are applied at each natural frequency (estimated from an initial ‣broadband test).The amplitude and phase of the forces are determined from the FRFs or else adjusted ‣iteratively until a normal mode is being excited (indicated by the forces and responses being monophase, with a 90 phase shift between excitation and response such that the structure behaves in essence like a single degree of freedom system). The mode shape is then measured and damping/modal mass estimated usually by varying ‣the excitation frequency around resonance. This approach is more suitable for studying important non-linear effects such as control ‣free- play, pylon stiffening, etc.

◦The identification of multiple-mode non-linear systems is a challenging area of research.•

Flight Simulator Test

A significant amount of testing is undertaken prior to the first flight to assist in Flight Control System •design via examination of the aircraft handling qualities.

Later on, simulators are used for in-service pilot training. •

The simulator is controlled to match the characteristics of a rigid aircraft dynamic (flight mechanics) •model.

Because of the impact of static aeroelastic effects upon the rigid body aerodynamic derivatives, •especially for large flexible aircraft, it is important to incorporate static aeroelastic corrections into the aerodynamic model used for the flight simulator.

It is also worth mentioning the ‘Iron Bird’ tests:•

include hardware for hydraulic/electrical systems, the ‘real’ flight control computers and ◦simulation of the natural aircraft to close the loop.

These are of most value in loads and aeroelastics issues by providing system transfer func- ◦tions and performance constraints for modelling flight control systems.

Structural Tests

Most of the structural tests (CS/AMC 25.307) are aimed at demonstrating the limit and ultimate load •requirements, and so are strength related;

Tests are not specifically related to dynamic loads except that certain critical cases may be dynamic •in origin and that excessive deformation should not occur.

The amount of testing required will depend upon the classification of the aircraft: •

the most extensive test program would be set in place for a new structure, with full scale ◦subcomponent (e.g. spar), component (e.g. wing) and whole aircraft tests carried out to limit

and ultimate conditions;

whereas, for a derivative aircraft, considerably less testing would be required. ◦

Flight Flutter Test

The uncertainty in the aeroelastic model used for flutter calculations, and especially the unsteady •aerodynamics, means that calculated flutter speeds will almost certainly be inaccurate to some extent, especially in the transonic region.

It is therefore a requirement of the certification process (AMC 25.629) to validate the flutter behaviour •and demonstrate freedom from aeroelastic instability over the flight envelope in a Flight Flutter Test (FFT).

On the basis of calculations, a nominal flight envelope is cleared to permit a first flight to take place. •

Thereafter, the FFT program precedes every other flight test at each flight envelope point because of •the safety critical nature of flutter. The basic FFT philosophy seeks to gradually extend the flight envelope by assessing the flutter stability of the aircraft at progressively increasing speed and Mach number.

It is normal to assess the flutter stability by identifying the frequency and damping of the complex/•damped modes of the aircraft at each test point. The allowable flight envelope is expanded from an initially agreed boundary by examining the results along lines of increasing EAS at constant altitude and lines of constant Mach number.

The procedure at each test point is:• (a) to excite vibration of the aircraft over the frequency range of interest and to measure its response, (b) to curve-fit the excitation and response signals in the time or frequency domain and so to identify the model parameters and (c) to determine whether it is safe to proceed to the next test point.

A variety of excitation devices can be used, namely:•

(i) control surface movement via stick/ pedal input or explosive charges, (ii) control surface movement via a signal from the Flight Control System, (iii) movement of an aerodynamic vane fitted to the aircraft flying surface or engine/store or (iv) inertia exciter mounted in the fuselage.

The aircraft response is measured using typically around 20 –100 accelerometers, far fewer than for •a GVT.

The most common excitation signals are pulse (via stick/pedal or explosive charge) and chirp (a fast •frequency sweep applied as a signal to the control surface, vane or inertia exciter).

Where excitation devices are available on both sides of the aircraft, excitation may be applied in or •out of phase in order to exploit symmetry/antisymmetry; doing this will simplify the analyses and provide results with more confidence.

Occasionally a random excitation signal is employed. Sometimes the response of the aircraft to •natural turbulence alone is used (but this is not recommended because the excitation is not ‘white’ and is not guaranteed to excite adequately all the modes of interest).

Each excitation sequence will probably only last a maximum of 60 s because of the difficulty of •holding the aircraft on condition, especially near to the limit of the flight envelope.

The test may be repeated and some form of averaging employed to improve the data quality.•

Once the test is completed, the results are processed on the ground, either during or after the flight. •FRFs are computed, or else the raw time data are used directly.

Time or frequency domain identification algorithms (i.e. curve-fits) are then employed in order to •identify the frequency and damping values for each mode in the data.

The process is very similar to modal testing but the levels of noise on the data are far more severe •since turbulence (an unmeasured excitation) is exciting the aircraft during the test. The test time is also limited.

The damping values may be compared to results from previous test points and the damping trends •for each mode extrapolated to allow progression to the next test point (i.e. by defining a permitted increment in speed or Mach number).

The test process is complete when the extrapolated damping values are still positive at a margin •(typically 15%) above the design dive speed for each Mach number.

These results are compared to the predictions from the aeroelastic flutter model and some basic •attempts may be made to reconcile any differences.

Any flutter problem (e.g. the anticipated flutter speed from test is too low) will require urgent design •action (e.g. additional mass balance) and could prove to be extremely costly.

Note that in AMC 25.629, the evaluation of any phenomena not amenable to analysis (e.g. buffet, •buzz, etc.) should be investigated during the flight test program.

Flight Loads Validation

In CS 25.301(b) it is stated that• ‘methods used to determine load intensities and distribution must be validated by flight load measurement unless the methods used for determining those loading conditions are shown to be reliable’.

The definition of such a flight loads test program is considered in AMC 25.301 and depends upon a •comparison of design features with previous aircraft (i.e. new features/configurations will require assessment), the manufacturer’s experience in load validation and proven accuracy of analysis methods, etc.

The aim of the flight loads programme is to demonstrate that the loads calculation and prediction •process produces reliable loads for the flight certification cases. In some cases where analytical methods are inadequate, such as for buffet, then flight loads are used for design purposes.

The predicted loads need to be validated by flight manoeuvres, both equilibrium and dynamic, and •ground manoeuvres.

Typically, load cases up to 85% limit load will be explored for separate symmetric and asymmetric •manoeuvres.

In particular, internal loads (namely bending moment, shear force, torque) will be required for •comparison with the model predictions at a number of spanwise stations on the major components (e.g. wings, horizontal tailplane, fin).

Since these internal loads quantities cannot be measured directly in flight, they must be inferred from •an array of strain gauge measurements.

Such a calibration relationship between internal loads and strain measurements may be obtained •from load tests carried out on the ground before first flight.

Once suitable calibrations have been obtained, then internal loads may be estimated under different •manoeuvres and compared to results of an onboard loads monitoring model based on a dynamic manoeuvre model. Any significant differences in results would be reconciled by adjusting this model.