Fretting fatigue crack nucleation analysis under a variable coefficient of friction

Abstract

Fretting fatigue is a phenomenon that occurs when two surfaces come into contact and experience oscillatory stresses and displacements of small amplitude. This can lead to a significant reduction in mechanical parts’ lifetime. The complexity of this type of problems relies on the many variables that can affect the results. One of its most important variables is the Coefficient of Friction (CoF), which has been observed to vary throughout a single fretting test. However, this variation is rarely considered in fretting fatigue finite element analyses. In order to analyse the effects of a Variable Coefficient of Friction (VCoF) on nucleation lifetime, a cylinder-on-plane finite element model is developed. To predict nucleation lifetime, two critical plane damage parameters, Fatemi-Socie (FS) and Findley parameter (FP), are used along with an evolving CoF during the first stress cycles. These parameters are evaluated by calculating initiation lifetime where the maximum value of the damage parameter occurs (hotspot method) and by the quadrant averaging method, because of its capability of predicting lifetime with relative accuracy and negative crack angles. To simulate the evolution of the CoF, a Python script that changes the value of CoF before each simulation is developed. This coefficient varies between 0.3633 and 0.65, following arbitrary points on the coefficient’s evolution curve. Once the simulation is concluded, the script evaluates the damage parameters and their equivalent lifetime throughout the contact surface and saves those values in an accumulation parameter that follows Miner’s rule. This process is repeated until CoF reaches its maximum value. The lifetime to nucleation is statistically analysed for all methodologies. The results suggest that, regarding the hotspot method, FP values remain within ± 3 N deviations from the experimental results, while FS results are nearer, within ± 2 N deviations. In addition, the quadrant method results remain at ± 3 N deviations, but providing more accurate results for the FP parameter. The application of a VCoF to these methods leads to less conservative and less deviated results, providing more accuracy for conservative approaches and more consistency for predictions in general. It is concluded that VCoF analysis is an interesting tool to determine more accurately nucleation lifetime for conservative approaches. Additionally, it reduces the predictions variation for differently loaded fretting setups regardless the desired approach.