Investigating the fatigue failure in five-axis machined components with multi-scale surface features using the Theory of Critical Distances and the FKM-Guideline

Abstract

Traditionally, casting has been used to manufacture components with free-form surfaces. However, the use of five-axis milling with ball-end cutters has become increasingly prevalent. This manufacturing technique allows for the utilisation of stronger billet materials in place of cast materials. On the other hand, it results in the formation of distinctive machining cusps on the surface of the component. Smaller scale machine marks appear within these cusps due to machine feed and tool cutting edges. Therefore, the multi-scale nature of such component’s surface roughness presents a significant challenge during fatigue failure assessment.

In this work, two fatigue assessment methods, the Theory of Critical Distances (TCD) and the FKM-guideline have been investigated to assess the fatigue failure of five-axis machined components. The FKM-guideline allows the use of FEA to quantify peak local stresses generated by cusps and applies empirically derived correction factors to assess fatigue life of components. TCD uses linear elastic stress distribution generated by the cusps, along with plain material and notched material fatigue data for fatigue assessment.

The FKM-guideline has been adapted, and its use has been extended to five-axis machined components to predict fatigue failure. The proposed model conceptualises cusps as stress-raising notches and the marks within cusps as surface roughness, characterised by the parameter R10z. Fatigue life assessment is conducted using stresses derived from a finite element analysis (FEA) model, while considering the effects of stress concentration, stress gradient, mean stress, and surface roughness. All predicted values (𝐹𝐾𝑀97.5%) are above the experimental life Ps = 97.5 % when the effect of surface marks within the cusps was not considered. The FKM methodology, in this case, overpredicts the fatigue life significantly. When cusps are treated solely as surface roughness within the conventional FKM framework, the resulting fatigue life predictions (𝐹𝐾𝑀97.5% significantly underestimated. In contrast, predictions (𝐹𝐾𝑀97.5% 𝐾𝑅,𝜎𝑐 ) are 𝐾𝑅,𝜎 ) that account for surface roughness within the cusps, while treating cusps as stress-raising notches, provide results that align more closely with experimental data but tend to overestimate the fatigue life.

The FKM method has been further optimised by incorporating 3D areal surface roughness parameters to investigate whether the FKM model can improve further. Results have shown that fatigue prediction (𝐹𝐾𝑀97.5% 𝐾𝑅,𝜎𝑆𝑧 ) improves when 3D surface roughness parameters are integrated into the FKM methodology. In this assessment, 3D parameter Sz gave the best correlation with experimental fatigue data Ps = 97.5 %.

TCD’s Point Method (PM) and Line Method (LM), has been adapted to model machining cusps as stress raisers. These adaptations have been integrated with the Medium Cycle Fatigue (MCF) procedure of TCD to predict the fatigue life of specimens machined using distinct tool sizes. The resulting predictions were found to be conservative. However, while comparing the predictions, TCD-LM method demonstrated superior correlation with experimental data as compared to the TCD-PM method. The TCD-LM method showed acceptable results for longer life specimens, however 40 % of data was still seen to fall outside acceptable ±2 scatter bands. TCD conservatism in results is primarily due to neglecting 3D stress raiser and plasticity effects. The effect of using Goodman, SWT and Walker mean stress models was investigated on predictive accuracy of the TCD method. Results showed Goodman mean stress correction gave better results as compared to SWT and Walker methods.

A comparative assessment of the FKM and TCD method's fatigue life prediction ability revealed that the TCD method generates conservative predictions, which could be preferred for safe life prediction. But TCD did not allow consideration of within the cusp minor surface marks. On the other hand, the FKM method allowed the application of more detailed fatigue assessments, by considering the effect of within the cusp surface marks. While this is the case, the FKM method overestimated fatigue life prediction, which could lead to premature failures.

Investigations in this work revealed that there is a substantial scope to develop and optimised these TCD and FKM models further to increase their accuracy. This has been presented and discussed in the concluding chapters of this work.

Date of Award	24 Oct 2025
Original language	English
Supervisor	Simon Barrans (Main Supervisor) & Karl Walton (Co-Supervisor)