Dynamic Structural Testing
Exciting a wing structure with oscillatory inputs is known as ground vibration testing (GVT) and is an essential preliminary ground test normally conducted prior to the beginning of flight testing. The main objective is to obtain the primary natural frequencies, mode shapes, and damping. To do this, the wing is excited near the root and aft of the main spar at a suitable strong point on the structure (such as at the torque reaction lug). This allows the shaker to excite both flapping and twisting modes of the wing at the same time. Accelerometers are fixed centrally near the tip and across the chord from front to back closer to the root so as to be most sensitive to the dominant flap and twist modes that need to be assessed. The system is then typically first driven with pseudo-random white noise, and a fast Fourier transform system is used to display the frequency response functions recorded by the accelerometers, suitably normalized by the excitation force transducer reading. This will reveal the various natural frequencies, and the widths of the response peaks show the damping levels.
To establish the mode shapes, the vibration signal is then changed to a pure tone at the frequency of interest, making sure that excessive motions are not stimulated by accident and the structure is illuminated with a powerful strobe. Stroboscopic illumination allows the tester to positively identify the associated mode shape for each peak in the frequency response spectrum. It is important to distinguish between the higher flap modes and the first twist mode: the first flap mode is usually obvious without stroboscopic lighting. This series of tests should confirm the values used in the previous aeroelastic analysis and thus remove any concerns over divergence, control reversal, or flutter. If the frequencies predicted by structural analysis are significantly different from those revealed by experiment, the analysis should be revisited to try and establish why. The most likely causes will be differing boundary conditions or the failure to include all the items contributing to the overall mass of the wing. These generally lead to the calculated frequencies being higher than those found from experiment. Errors of 10-20% are quite common and should be allowed for when predicting flutter, divergence, and control reversal onset speeds by adopting suitable margins of safety.
Figures 16.16 and 16.17 show Decode-1 on-ground vibration test, while Figures 16.18 and 16.19 show the frequency response plots for two accelerometer positions with the cursors set for the two modes of interest. The relevant natural frequencies are seen to be 5.78 and 59.25-60.00 Hz. These values are slightly lower than those predicted by the FEA model
Figure 16.16 Ground vibration test of a Decode-1 wing showing support and mounting arrangements.
Figure 16.17 Ground vibration test of a Decode-1 wing ((a) accelerometer on starboard wing tip: (b) shaker and force transducer near wing root).
Figure 16.18 Frequency response from ground vibration test of a Decode-1 wing: accelerometer on port wing tip and cursors on first flap mode.
described earlier. This is generally found to be the case as noted above: here, the errors are between 6% and 8%.
Note also that the resonance at 3.1 Hz is a combined flap and rigid-body roll mode that occurs because of the way the model was mounted. This is easily distinguished because the wing tips
Figure 16.19 Frequency response from ground vibration test of a Decode-1 wing: flapping mode accelerometer placement (upper) and twisting mode placement (lower), cursors on first twist mode.
move in anti-phase with each other. The mode at 7.3 Hz is caused by flexure of the entire tail in opposition to motion of the wing and is again easily identified by its mode shape: this mode is also seen in the FEA model. That at 33 Hz is the second flap mode that occurs before the first twist mode, as was predicted by FEA and which can be identified by stroboscopic illumination. The low-frequency rigid-body modes can also be eliminated from consideration by observing that they shift significantly if the airframe supports are changed and tests repeated: for example, by testing with the aircraft sitting on its own suspension and then by supporting it on mounting blocks or from a soft hanging suspension. The coherence values in the response plots confirm that nonlinear and signal-threshold effects are not significant at the resonant frequencies, being at 0.998 or better at the frequencies of interest. Tests should also be carried out on both wings to check that natural frequencies do not differ appreciably; if they do and the measured wing masses are similar, this can point to deficiencies in the manufacture that should be addressed before flight testing.
In Figure 16.19(upper), the accelerometer is placed at the wing tip directly above the main spar, while in Figure 16.19(lower) the accelerometer is placed at half span and on the trailing edge. Thus the upper set of results accentuate flapping mode responses, while the lower focus on the twisting mode. The magnitudes of the two responses can be seen to change by around 6dB between the two plots in the way expected, given the accelerometer placements.
The first flap and twist modes have damping ratios of 0.026 and 0.05-0.1, respectively. The damping stems mainly from the foam-plus-cladding nature of the structure, which has quite good intrinsic damping characteristics compared to pure CFRP or glass-fiber reinforced plastic (GRP) structures.
-  We once lost a student aircraft in flight because of a defect in a main spar, which would have been revealed by suchtests had they been carried out.
-  The width is measured as that required for the height of the peak to be reduced to 0.7071 of its peak value, or by3 dB, and the damping ratio is then given by ? = 0.5 Дф/ф: here, the width of the flapping and twist modes are 0.3 Hzand 6-12Hz - the twist mode resonance being somewhat difficult to exactly define.