Health monitoring of a complex structure using a sparse array of modally selective transducers

Abstract

We describe a novel method of monitoring the structural health of a lattice-like structure using a permanent sparse array of structural-acoustic modally selective transducers. The transducers are used to measure transfer functions across the structure. Changes in these transfer functions can indicate changes in the structure, including corrosion, cracking and physical damage. Combining information across the set of transfer functions can allow the location of a change in the structure to be pinpointed. The method makes possible low-cost, automatic monitoring of large complex lattice-like structures.

In this paper, we report results of trials on a scale model of an offshore jacket (for example, the cross-braced structure that supports an oilfield production platform in the North Sea). Results demonstrate quite accurate location of damage at two well-separated points on the structure, using two different arrangements of acoustic sources and sensors. The first arrangement—in which sources are at the top and sensors at the bottom of the structure—showed limited vertical resolution of damage location. The second arrangement—with both sources and sensors at the top—shows much better vertical resolution. Best results would be obtained by deploying sensors at both top and bottom, but this is probably impractical on jackets.

The main limitation was signal drift due to both electronic drift and temperature drift in the structure, which masks small changes in transfer functions caused by structural damage.

Keywords:

• Structural monitoring
• Structural acoustics
• Structural dynamics
• Sparse array
• Guided waves

Acknowledgements

The authors would like to acknowledge the encouragement and financial support of the UK Health and Safety Executive.

Notes

^† A “path” here means any path that can be traced through the structure from transmitter to receiver via which structural vibrations can travel. Since almost all paths contain changes in mechanical impedance, any signal travelling along them will undergo multiple reflections, producing a reverberating signal that continues until absorption and radiation attenuate the vibrations.

^† In practice, the wave velocity used to construct this table was estimated using the onset times of the measured transfer functions. This gave a velocity slightly lower than that measured on a single bar of polycarbonate stock, and we concluded that this was an effect of the joints. We did not attempt to model the dynamics of joints in detail.

^† The laboratory equipment used to digitise the signals in early experiments suffered from limited stability of the sampling interval. This made it desirable to keep the signal bandwidths as low as possible. However, we also wanted to obtain good resolution in locating points of damage, which pointed to using wave pulses which were physically as short as possible. Thus, there was an advantage to using a material with a lower acoustic velocity.