Vibration Testing
Whenever a component, assembly or structure is subjected to vibration, the risk of failure is increased as individual resonances can combine, or structural imperfections are amplified, or fasteners not tightened correctly start to loosen. The more severe the vibration, the more likely that failure will occur, and some of the most severe vibrations can be found during a rocket launch into space. The failure of a single component during launch could essentially see the end of a whole mission that may have taken many thousands of man-hours to reach the launch stage, so one of the most vital tests performed in achieving space-flight qualification is the vibration test.
The vibration test actually begins long before any testing takes place. The mathematical model is first analysed to determine eigenmodes with their percentage mass participation, and also stress analysis performed to determine what levels of stresses are acceptable during external loading. Once this has all been completed, then the test proper can begin.
It should be noted here that with space hardware, once a test regime has begun, you cannot alter the Device Under Test (DUT) in any way, and that means opening it to take a quick look if something seems wrong as this will invalidate all the previous tests performed!
The following is a much simplified explanation of what happens during the vibration test!
Low Level Sine Sweeps (LLSS), or Resonance Searches, are, as suggested, a sine sweep from 5 - 2,000Hz at 0.2 or 0.5g at 2 Oct/min. The LLSS allows you to 'see' inside the DUT as it shows all the individual resonances which can then be compared to the modal analysis to see exactly which components are being excited. This is the litmus paper test as it allows us to make comparisons throughout the test.
The part of the test that stresses the DUT is the Random Vibration, which is white noise that fills an envelope. It is performed in all three axes, starting at -9dB, then -6dB, followed by -3dB. After each level, the Grms values at each accelerometer are calculated up to determine the expected full level. If this value starts to increase or shows a figure, multiplied by 3 for 3-Sigma, higher than the stress analysis allows, then the testing is stopped. If all is ok, then Full Random is applied.
After each axis has been completed, the LLSS is repeated and compared to the initial LLSS in that axis. The vibration test is deemed a success if none of the frequency amplitudes have increased by more than 20%, and frequencies changed by less than 5% on modes with an effective mass greater than 10%. Or is it that simple?
Put yourself in the situation I was in. I'm performing a vibration test on a mechanism that will be launched into space as it's the Flight Model. All the Random Vibration runs have been performed in one axis, and I am analysing the Low Level Sine Sweeps, but there's a frequency and amplitude shift that might be a cause for concern, but then again, it might not. The options are to either stop the test and open up, which invalidates all the previous tests, or convince ESA that we should continue because what we might be seeing is a small frequency shift that now highlights a peak that was only partially seen before because of the sampling rate. Make a wrong call and the whole 2 Billion Euro program could end in failure during the launch. What would you do?