0 1 2 3 4 0 0.5 1 1 True Strain [-] Relative Radius [-] R = 0.45 mm R = 1 mm R = 5 mm smooth .5 Figure 5. Relative radius predictions according to Rice and Tracey, ART = 3/2. Influence of the Strain Rate The damage predictions presented in this work do not include a dependence of the strain rate directly. This can be seen from the formulation used for local failure prediction, Eqn. 2. Implicitly this mechanism is accounted for by strain rate dependence of the flow stress. Experimental results confirmed the choice of this relatively simple approach. No significant dependence of the fracture characteristics has been observed from the tensile tests at different rates of strain. Elongation, reduction of area at fracture, and fracture surface showed no remarkable changes at elevated strain rates. A minor reduction of strain from area reduction might be concluded. The influence of the strain rate on the local damage prediction in quasi-static analyses has been investigated as well. In these analyses the mass inertia influence has been neglected. In these analyses the strain rate influences the flow stress and will result in strain rate dependent damage evolution. However, physical processes that might influence the void growth on a different basis than the influence of the strain rate on the flow stress have not been taken into account. Reduction of strain at failure has been predicted from the numerical results, applying the same critical damage value for all strain rates. For an increase of the strain rate of 1.105 s-1 the failure strain reduces about 0.1 for the smooth specimen and about 0.06 for all notched geometries. These values agree with the experimental observations. Introduction of a Length Parameter Failure predictions using a characteristic volume might improve the analogy between experimental results and numerical computations. This volume represents a certain process zone in which failure takes place. The characteristic length of the zone can be related to microstructural characteristics, such as the average inclusion size or grain size. The finite element model has been chosen is such a way that the element length in front of the notch tip is equal for all notched specimens. Damage has been evaluated as the average value of each element. As the formulation for strain calculation uses an updating of the incremental displacements (Updated Lagrange), the elements deform in a similar way as the material structure. This can be interpreted as to account for the changing inter-inclusion spacing at proceeding deformation. The element length has been chosen in such a way that the model size would not be excessively large. This element length is approximately 210 µm and might be related to the average inclusion distance of the manganese sulphides, which has been estimated to be 310 µm. The critical damage value from a smooth tensile specimen changes not significantly using the constant element size approach, as the gradient in damage is small in the specimen centre. A calculation has been performed applying constant element size with similar dimensions as has been used for the notched specimen. From this calculation a damage value of 3.7 has been found. From the finite element results can be concluded that fracture initiation in the notched specimens takes place at lower damage values compared with the smooth tensile specimen. Taking a closer look to the damage curves, the critical value of the relative radius at failure initiation is approximately 2.5 ± 0.35 for all notched geometries. For the smooth specimen this relative void radius corresponds to a strain of 1.2 ± 0.1. Interrupted tests have been
RkJQdWJsaXNoZXIy MjM0NDE=