Enhancing impact localization from fluid-pipe coupled vibration under noisy environment

Abstract

Big impacts from pressure transients are known to be major contributors to pipeline failures, and even small impacts have the potential to result in pipeline deterioration. The effects of these impacts on a pipeline are not disclosed in detail and are hard to evaluate completely by theoretical modeling or numerical simulation. The impacts excite cylindrical structures containing liquid, which results in the development of wave propagation along pipelines. In particular, quasi-longitudinal waves are known to be developed by fluid-structure interaction during propagation. However, impact signal detection needs great care because extreme noise may corrupt the signal. An enhanced wavelet-based approach is proposed to detect and localize the impact source. The method makes use of continuous wavelet transform and band summation within a band of interest along scales to enhance time-difference detectability. In addition, a noise reduction algorithm intended to remove burst noises, in practice, frequently contaminates the impact signals of interest. Experimental results from a water supply network under operation demonstrate that the proposed approach is able to suppress noise and successfully reveal the impact location. The proposed approach provides a more precise and robust way to localize impacts missed by the conventional cross-correlation algorithm.