Detection of Multiple Cracks in Gears using Transmission Error

Geared transmission systems are commonly used in various industries; however, the gears themselves are prone to failure due to mechanisms such as fatigue and abrasive wear. One of the most serious gear faults is fatigue cracking at the root of the tooth, where bending stress is high from the meshing forces during tooth engagement. Once initiated, these cracks propagate under repeated cyclic loading, leading to breakage of the tooth and, ultimately, the complete failure of the transmission system. Techniques to diagnose root cracks from measured vibration signals are now well established and are generally based on the detection of a sudden drop in the ‘meshing stiffness’ that occurs when the cracked tooth comes into mesh. Such techniques can be separated into two main categories. The first ‘models’ the phenomenon as a sharp modulation of the regular gearmeshing pattern, and is thus based on the identification of sidebands spaced at shaft speed around the gearmesh frequency in the vibration spectrum. The second instead sees the meshing of the cracked tooth as a sudden deviation or impulse in the meshing force, triggering an additive impulse response, with the response characteristics determined by the system’s dynamic properties and a repetition rate dictated by machine speed. Yet, while both approaches have been used successfully to diagnose gear cracks, vibration-based techniques can be strongly affected by operating conditions, and they are generally unable to indicate the severity of cracks. Furthermore, limited work has been done on multiple crack detection. This paper uses transmission error as an alternative to vibration analysis in detecting multiple cracks. TE is found to be effective in detecting the position of multiple cracks and can indicate the different severity of the cracks.