Search results

The main goal here is to optimise the finite element mesh used to predict plasticity induced crack closure (PICC). A numerical model was developed for a M(T) specimen made of 6016-T4 aluminium alloy. The parameters studied were the size of most refined region perpendicularly to crack flank (ym) and along propagation direction (xr), the size of finite elements near crack tip (L1) and the vertical size of refinement close to crack flank (yA/B). A maximum size of about 1.3mm was found for ym, but a smaller value has a limited impact on PICC. An analytical expression was proposed for xr, dependent on δK and Kmax. An optimum value seems to exist for L1.

Compressive stresses play an important role on tension-compression fatigue which can be attributed to plasticity induced crack closure (PICC). The objective here is to study numerically the effect of compressive stresses on PICC and to discuss the applicability of PICC to explain the effect of negative stress ratios on fatigue crack growth rate. The compression produces reversed plastic deformation at the crack tip, reducing linearly the crack opening level. The incursion to negative stress ratios did not produce sudden changes in the behavior of PICC and no saturation with the decrease of minimum load was observed for δKeff. Crack closure was able to collapse da/dN-δK curves with negative stress ratios, indicating the applicability of the crack closure concept to explain the effect of negative R. The analysis of crack tip plastic strain range with and without contact of crack flanks confirmed the validity of crack closure concept.

The mean stress has a significant effect on crack propagation life and must be included in prediction models. However, there is no consensus in the fatigue community regarding the dominant mechanism explaining the mean stress effect. The concept of crack closure has been widely used and several empirical models can be found in literature. The stress ratio, R, is usually the main parameter of these models, but present numerical results showed a significant influence of Kmax. A new empirical model is therefore proposed here, dependent on Kmax and ΔK, with four empirical constants. The model also includes the effect of material's yield stress, and two additional parameters were defined to account for stress state and crack closure parameter. A comparison was made with Kujawski's and Glinka's parameters, for a wide range of loading conditions. ΔKeff lies between Kujawski's and Glinka's parameters, and some agreement is evident, although the concepts are quite different. The crack opening model was applied to literature results and was able to collapse da/dN-ΔK curves for different stress ratios to a single master curve.