Gear diagnostics - Fault type characteristics

Abstract

To date, the majority of existing Condition Indicators for gears are based on various statistical moments of a recorded time history. A supplementary analysis proposed in this study, shall suggest an approach that may, in the future, enable the identification of faulty gearwheel and possibly fault type in the system. In this work, a combined analytical and empiric approach is applied. This approach is based on the assumption that reliable dynamic models can be utilized to predict the effects of faults on vibrational patterns. Dynamic model generated signatures are used to verify experimental findings. Moreover, discrepancies between simulated and actual results, combined with understanding of the assumptions and omissions of the model, are helpful in understanding and explaining the experimental results. A spur gear transmission setup was used for experiments, along with an electric AC motor and a friction belt loading device. The experimental runs were conducted at varying speed settings. Two types of faults, a tooth face fault and a tooth root fault, were seeded in the experimental transmission and into the model. The effect on extracted signal features is examined. The purpose of this study is to evaluate fault detection capabilities of proposed diagnostic tools at the presence of two seeded faults of varying severity, verified by a dynamic model. Observed differences between examined fault types and their manifestation will be discussed. A basis for future work on prognostics capabilities is laid by a varying degree of tooth root fault.