Analysis of Ball Bearing Defects in Synchronous Machines using Electrical Measurements
MetadataShow full item record
Rolling element bearings are used in most electrical machines, especially for small and medium size applications. Under non-ideal operating conditions, ball bearing condition degrades by fatigue, ambient vibration, misalignment, overloading, contamination, corrosion from water or chemicals, improper lubrication, shaft currents and residual stress left from the bearing manufacturing process. All of these conditions eventually lead to increased vibration and acoustic noise during machine operation which at some point in time results in unexpected bearing failure. Over the years, a great number of publications have been devoted to the detection of mechanical faults, including rolling element bearing defects and torsional defects, in electrical machines based on Electrical Signature Analysis (ESA). It has been observed that these faults can affect either the stator to rotor air-gap distribution or the running speed of the machine, which can be reflected in the signature of the electrical signals. However, the physical link between the mechanical degradation and the electrical signature is still not explained well. A multi-physics model is developed by joining the detailed mechanical model of a rotor bearing system and the electrical model of a synchronous machine in this research. This combined model is capable of describing the transmission of information originating from bearing faults and their impact on the variations of the measured electrical signals. The electrical machine model is developed based on winding function approach and its validity is demonstrated by a more accurate Finite Element Method (FEM) model. The mechanical model consists of a high fidelity rotor-bearing system with detailed nonlinear ball bearing model and a flexible finite element shaft model. It is validated using the housing vibration data collected from some experiments. Generalized roughness bearing anomalies are linked to load torque ripples and airgap variations, while being related to current signature by phase and amplitude modulation. Considering that the induced characteristic signatures are usually subtle broadband changes in the current spectra, these signatures are easily affected by input power quality variations, machine manufacturing imperfections and environmental noise. In this research, a new algorithm is proposed to isolate the influence of the external disturbances of power quality, machine manufacturing imperfections and environmental noise, and to improve the effectiveness of applying the ESA for generalized roughness bearing defects. The results show that the proposed method is effective in analyzing the generalized roughness bearing anomaly in synchronous machines. Furthermore, the electrical signatures are analyzed in a synchronous machine with bearing defects. The proposed fault detection method employs a Zoomed Fast Fourier Transform (ZFFT) and Principal Component Analysis (PCA) and it is also tested on the available experimental data. The results show that amplitude induced electrical harmonics are related to the level of vibration, and the electrical signatures are affected heavily by other variables, such as power quality and load fluctuation. The proposed method is shown to be effective on detecting generalized roughness bearing defects in synchronous machines.
Wang, Tengxi (2017). Analysis of Ball Bearing Defects in Synchronous Machines using Electrical Measurements. Doctoral dissertation, Texas A & M University. Available electronically from