In Figure 2 the one direction is parallel to the rolling direction and shows the average curves for 4 experiments. The second direction is perpendicular to the rolling direction of the plate material and the markers shown in the figure represent only one experiment. To compare these results in both directions, the scatter in measurements for the parallel direction has been presented by plotting a few raw data points of the flow-stress according to Bakker into the graph. These points indicate that no remarkable differences were found between both plate directions. Only the Bridgman flow-stress and Siebel flowstress deviate a little from the parallel direction. As the axial mean stress and Bakker flow-stress are practically identical and have been derived from load and diameter data only, one can conclude that the ratio a/R is slightly higher in the direction perpendicular to the rolling direction. Flow Curves at High Strain Rates For the experiments performed at high strain rate the results showed considerable scatter, as well the strain measurements from the camera images, as the load measurement from the piezoelectric signal. The scatter increased with increasing strain rate as the testing velocity largely influenced resonances. The failure strain appeared to be rather independent of the strain rate, a small reduction of failure strain in the rolling direction was found, yielding a failure strain of 1.3 and 1.4 at engineering strain rates of 60 s-1 and 200 s-1, respectively. The correction procedures according to Bridgman [2] and Siebel [3] have been applied. However, the Bakker procedure showed some numerical problems, and an alternative approach has been applied to derive the flow stress at high strain rate. A finite element simulation has been performed using a preliminary flow curve. The flow-curve assumed did not yield to accurate similarity between numerical simulation and experiment. One could observe a premature localisation and deviation of axial stress compared with the experimental data. For this reason, the initial part of the flow-curve up to a strain of 0.2 has been adapted to prevent necking in an early stage of deformation. A simple powerlaw description gave satisfactorily results. For strains larger than 0.2 the coefficients have been adapted to fit the mean axial stress level from the experiment. Table 1 shows the values of the coefficients used in the powerlaw flow-curve. This resulted in a proposed flow curve as has been shown in Figure 3, engineering strain rate is 5.7 10+1 s-1. Application of this flow-curve resulted in a better agreement between simulation and experiment, although the total elongation still has been underestimated by 10%. 0 250 500 750 1000 1250 0 0.5 1 1.5 True Strain [-] Stress [MPa] true stress (exp) true stress (FEM) Bridgman (exp) Siebel (exp) flowcurve Figure 3: Flow-curve for FEM – simulation that fits experimental curve for the true stress at an engineering strain rate e = 5.7 10 & +1 s -1
RkJQdWJsaXNoZXIy MjM0NDE=