Prediction of thin-rimmed gear crack propagation from a factorial design approach

Abstract

Gear failure involving bending fatigue can have catastrophic consequences depending on the propagation path direction. Therefore, anticipating and preventing eventual critical fracture are crucial at the design stage. However, none of the methods available can give rapid and quantitative evaluation of gear fatigue crack evolution. Aiming to provide fast predictions of crack propagation paths, this paper proposes a factorial design approach for gear bending fatigue simulation. Six parameters related to gear geometry and initial crack configuration are considered in this study. Factorial design experiments are numerically conducted with an efficient 2D boundary element model assuming linear elasticity. Then, bending fatigue damage is modelled using polynomial functions. The resulting prediction model can instantly establish the crack trajectory in thin-rimmed gear for any cycle numbers. Application of the approach is illustrated by several case studies, while its precision and reliability are demonstrated through an exhaustive validation procedure.