Contact Fatigue Analysis of Helical Transmission Gear Using Finite Element Method with Material property Prediction by Artificial Neural Network Model
No Thumbnail Available
Date
2020-02
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Gear is the most essential element in power transmission system. Helical transmission gear
can operate at high speed with large load carrying capacity. Due to this high contact stress is
created at the mating surfaces. One of the main gear tooth failure type is contact fatigue failure
due repetition of high contact stresses. In addition to design aspects, two important areas need
to be addressed in order to enhance helical gear damage due to contact fatigue; improvements
of material and enhancement in heat treatment. But it is very difficult to develop a complete
theoretical/analytical model to improve the material property and heat treatment. In addition,
to perform those enhancements, it needs an experimental work. In this study prediction of
mechanical property of helical gear material using artificial neural network (ANN) and
analyzing the contact fatigue of the predicted materials has been performed. After training
the network, different performance measurements of the neural network accuracy was taken
and prediction of the new concept (mechanical property) was performed. From five candidate
materials, concept one was selected. By using the developed mechanical properties, contact
fatigue was analyzed using AGMA standard and Finite Element Method. The results
indicates, the fatigue life is infinite until the contact stress reach 959.7 Mpa. But beyond this
contact stress, the fatigue life is limited and decreased. The comparison of contact stress by
using AGMA and Ansys for the predicted material using ANN has shown and an error of
4.46 % and below was obtained. The material has best performance until the applied
tangential load reaches 2000 N, because for applied tangential load of 2000 N, the factor of
safety for AGMA as well as Ansys is greater than one. This indicates that, it is selective
technique to predict the mechanical properties of materials using ANN model, when there is
limited condition to use experimental investigation, because ANN simulates any correlations
that are difficult to describe using physics based models.
Description
Keywords
Artificial Neural Network, Contact stress, contact fatigue, fatigue life, Finite Element Method