Fatigue life evaluation method of metallic components under random loads based on a new strain-life curve

•In random load conditions, the real service life of parts and components was verified by road simulation durability test, the prediction risk of fatigue life based on ε-N curve is evaluated through the quantitative analysis approach.•A double-parameter Coffin-Manson model is proposed. Its advantages are that the damage related to mechanical strength or machining process of parts and components can be corrected by the part ε-N curve.•The double-parameter model can be were calibrated further by random load factor β, thus the fatigue damage of random loads can be conveniently and precisely calculated by using one parameter.•The model has strong openness and expansibility in aspects of ε-N curve parameter, it can also be applied to quantify the material dispersion. The fatigue life prediction methods of constant-amplitude load have been widely studied. However, a simple and mature random loads fatigue model is more worth exploring. In this paper, the battery box brackets were taken as the study case, the road simulation technology and statistical method are adopted to analyze the prediction accuracy of random loads which is based on ε-N curve and damage accumulation law, a fatigue damage calculation procedure of random loads was established. Firstly, both the road simulation and constant-amplitude experiments of bracket were conducted respectively on the hydraulic servo rig. Simultaneously, the strain loads were collected and the part ε-N curve was constructed. In accordance with the material and part ε-N curve, an improved double-parameter (u and v) Coffin-Manson model was proposed and a fatigue ε-N curve was plotted. Then the damage originated from machining process and geometrical uncertainty was corrected by u and v. Under random load conditions, u and v were calibrated furtherly by random load factor β. Finally, the damage of the brackets was recalculated and the result has a high precision. The double-parameter model coupled with β has strong adaptability to most fatigue uncertainties. A leaf spring experiment was also performed to verify the generality and validity of the model.