Prediction of the Fretting Fatigue crack nucleation endurance of a Ti-6V-4Al/Ti-6V-4Al interface: Influence of plasticity and tensile/shear fatigue properties

摘要

Fretting fatigue is a critical loading that appears on many structures such as the blade/disk contact of aircraft engines. It is characterized by small displacements between two bodies, one of which has an applied bulk stress. This phenomenon has been studied for a long time and induces damages which critically reduce the fatigue resistance. Fretting stressing are multi-axial and above all characterized by very severe stress gradient conditions. Besides, significant contact plastic deformations can be activated for LCF conditions which again complicate the predictions. To address this aspect a combined experimental - modeling analysis has been developed on a well known a Ti-6V-4Al alloy. Using an original double actuator fretting fatigue apparatus, the fretting fatigue cracking endurance of a cylinder/plane contact from 10E4 to 10E7 cycles was determined keeping constant the partial slip fretting loading (i.e. Q/P=0.32) and varying the fatigue stressing ( 700 MPa < σ_(max)<50 MPa, R=0.01). This research shows that the fretting fatigue crack nucleation endurance (b_(CN)=70 μm) can be predicted using a basic Crossland's stress invariant multi-axial fatigue description if the cyclic elasto-plastic (ep) response of the material is conveniently integrated in the FEM simulation, the stress gradient effect is taken into account by considering a stress averaging strategy defined on a l_v = 37 μm cubic size previously calibrated from plain fretting experiments and finally if the ratio of tensile and shear fatigue limit is not assumed constant but expressed as a function of the loading cycles (σ_(-1)N)/τ_(-1)(N)). Using this complete analysis the crack nucleation endurance from 10E4-10E7 cycle is predicted with an error smaller than 10%. A parametric study shows that if the σ_(-1)/τ_(-1) ratio is assumed constant, defined at the fatigue limit conditions, the endurance predictions in LCF and most of the HCF domains are dangerously non conservative. Alternatively the application of plain elastic stress description leads to over conservative predictions when significant plastic accommodations are activated (N < 2 10~4 cycles).