In recent pilot studies we have started investigating how to possibly use measured flexural (i.e. transverse/bending) vibrations, induced by bolt-tapping, to estimate bolt tightness. Some of the vibration features we investigated showed strong correlation with bolt tightness. For example, the lowest natural frequency, as estimated from the vibration signal, is close to being proportional to bolt tension, except at very low tensions (Fig. 1(d)). To obtain an estimate of bolt tightness this way only requires a period of time of the order of a second where the user taps the blot with a light hammer, whichtriggers measurement and data processing. However, experimental results revealed that this technique encounters twoproblems, for it to be presently useful in real applications: First the variability in results is too large, about twice that for a torque wrench. Figure 2(a) shows that the slope of the curves varies for separate experiments with similar increasing bolt tension, which suggests that the quantity ω/β is an unreliable feature for estimating bolt tension. A second fundamental problem is that the relative change in natural frequency is approximately proportional not only to bolt tension, but also to slenderness ratio. Thus, if only the natural frequency feature were to be used for estimating bolt tension, accuracy will drop off for the short and thick bolts that are often used in critical joints.
展开▼
