Detection of rub in rotating machineries by Wavelet analysis of vibration data

Rubbing between the surfaces of the rotor and stator in a rotating machine can be detected and quantified by analyzing the vibration signal from the machine. Rubbing leaves its signature in both the high frequency and the low frequency region of the signal which can be detected using Wavelet analysis. The rub-related component in the high frequency region of the signal cannot be used to detect rub since it can be masked by high amount of noise present in industrial environment. The low frequency component is less vulnerable to noise and can be efficiently separated from the signal using Discrete Wavelet Transform. Once separated, the rub-related component of the signal can be quantified to measure the amount of rub present in the rotating system. This paper explains and illustrates the use of Discrete Wavelet Transformation to detect and quantify rub-related faults in rotating machineries. A simulated case study using the Jeffcott rotor model and a pilot plant experimental case study are also presented to demonstrate the efficacy of the proposed technique.