Random Vibration Testing Mirrors Real-World Effects
When specifying and conducting random vibration tests on automotive components, you no longer need complicated PSD (power spectral density) or ASD (acceleration spectral density) graphs.
Wayne Tustin, Equipment Reliability Institute, Santa Barbara, CA -- Test & Measurement World, 2/15/1999
Auto makers typically acquire and store vibration data, and later extract and analyze it, using the method illustrated in Figure 1. They mount an accelerometer or other sensor on an automotive component such as a seat, instrument, instrument panel, or window. A developmental vehicle (which may contain numerous such components) performs appropriate maneuvers on smooth and rough roads or test tracks, at various vehicle speeds and engine speeds.
Meanwhile, onboard or remote equipment gathers and stores data about how each component responds to road and engine vibration inputs. Later, a Fourier-transform spectrum analyzer derives from that data a spectrum such as that shown in white in Figure 2. Figure 2 also shows two historic versions of "wrapping" a test spectrum around that data. The analog shaker controls used in the 1960s and 1970s could not reproduce spectra much more complex than the approximation shown in red. Similar spectra still appear in test and screening programs, even though digital shaker controls available since the 1980s permit much more complex test spectra, such as the one shown in yellow, which has steeper slopes and more break points.
The test spectrum indicated in yellow is an improvement over the one in red. (It more closely approximates the white field-data spectrum, and the reduced area beneath it implies a reduction in RMS acceleration, so a given shaker can vibrate a heavier load.) It still fails to adequately simulate the field environment, however, because of the averaging that is a necessary part of transforming field data into a test spectrum. Averaged spectra—the only kind you normally see in test reports and test specifications—may fail to give a complete picture of the dynamics of the field environment. Take your own experience driving a car. You’ll agree that most road surfaces (and thus the ride’s statistical properties) continually change. When a motorcycle, car, bus, or truck traverses a similarly changing road or test track, each onboard accelerometer develops an electrical signal representing the occasional bump or pothole. An onboard or telemetered recorder stores the peak voltages representing the effects of bumps and potholes on the component. Averaging the Samples Is that bad? Not when studying some kinds of vibration—for example, in most rotating machinery and much routine aircraft flight, successive samples change little if at all. But for automotive test, averaging hides some of the most important data: the effect on the vehicle of the really severe peaks and valleys that can reach 12 or more standard deviations. Today, you can perform tests with greater realism and eliminate several processing steps. You no longer need to generate spectra, average them, plot the average, sketch straight-line approximations to form a test spectrum, and adjust the shaker controller to match that spectrum. Instead, you can stream data directly from your field data recorder into your shaker control computer’s hard disk. From there it will go to the power amplifier that drives the shaker—not directly, but rather through a shaker compensator (Fig. 3).
Shaker Compensation Adjusting the shaker compensator (Fig. 4) is simple. Attach an accelerometer at the same location on the test component where it was mounted during field recording, as Figure 4 shows. (Testing with the control accelerometer mounted on the component being tested is called "response-controlled" testing. That contrasts with the more common "input-controlled" testing, where the control accelerometer is mounted on the fixture that drives the component being tested.)
Drive the shaker system with a sweeping sine wave. The shaker compensator continuously adjusts its magnitude to maintain 1 g RMS at the accelerometer, storing the inverse characteristic (Fig. 5) of the shaker plus the fixture plus the test article, ready for use as shown in Figure 3.
The RMS control acts like the loudness control of your radio or music system. It increases or decreases intensity without otherwise changing the accuracy of the reproduced field environment. You can use it to experimentally determine the effect (upon the test article) of greater or less severe dynamic inputs. You can explore tradeoffs between severity and life expectancy experimentally. In the test lab you no longer need to understand or to explain (to customers) the term "power spectral density" or the alternate terms "acceleration spectral density" or "auto-spectral density." You won’t need to explain the rather weird units of g2/Hz, either. Designers will probably continue to use spectra and those terms; a design goal is to place resonances of new products in those frequency regions where vehicle vibration is least severe. Testing to a Spectrum
Certainly you will have some customers who won’t consider any changes and who will insist on following the present test plan. "We’ve always done it this way!" is the unfortunate reaction of some individuals and organizations. But progressive customers will listen and thank you for educating them. Perhaps the next test will be more realistic, employing the techniques outlined above. T&MW Wayne Tustin is the founder and president of Equipment Reliability Institute, Santa Barbara, CA; 805-564-1260. tustin@equipment-reliability.com Acknowledgement |
