Fatigue life estimation of cast aluminium alloys considering the effect of porosity on initiation and propagation phases

• Published in: International journal of fatigue 2019-08, Vol.125, p.468-478
• Main Authors: Hidalgo, R., Esnaola, J.A., Llavori, I., Larrañaga, M., Hurtado, I., Herrero-Dorca, N.
• Format: Article
• Language: English
• Subjects: Aluminum base alloys; Cast aluminium alloys; Casting alloys; Casting defects; Crack initiation; Crack propagation; Die casting; Fatigue failure; Fatigue life; Fatigue limit; Fracture mechanics; Fracture surfaces; Low pressure; Materials fatigue; Metal fatigue; Model accuracy; Phases; Pore size; Porosity; Pressure casting; Pressure die casting; Propagation
• ISSN: 0142-1123; 1879-3452
• DOI: 10.1016/j.ijfatigue.2019.04.006

•Fatigue performace of cast components is mainly affected by porosity.•An alternative fatigue procedure that accounts the effect of porosity was developed.•The model presented an average logarithmic error below 2% and absolute below 16%.•The required porosity parameters can be obtained by computed tomography.•In high series sectors, a statistical store with fracture surface data could be used. Porosity is one of the main parameters in the fatigue life of cast alloys as it affects both crack initiation and propagation phases. However, most fatigue assessments present a lack of accuracy as they only consider the effect of the initial pore on the fatigue limit. Therefore, in the present study, first, the effect of the main porosity characteristics on the crack initiation and propagation phases is evaluated. Then, an alternative fatigue assessment model is proposed and implemented for Low Pressure Die Casting (LPDC) A356-T6 aluminium alloy. The model accounts for the effect of the initial pore size and distance to surface on the initiation phase as well as fractographic porosity and average pore size on the fracture surface on the propagation phase. Finally, the model was validated by means of experimental tests. Results have shown that the proposed model is able to qualitatively predict the scatter trend of samples tested at the same load. In addition, the model presents high quantitative accuracy in life estimation with an average logarithmic error below 2% and an absolute average error below 16%.