The influence of plastic strain ratios on the numerical modelling of stretch forming

The orthotropic response of a 2024-T3 aluminium alloy was characterised through a series of uniaxial and biaxial tensile tests. The empirical information obtained from these tests was used to calibrate a non-quadratic orthotropic yield criterion for specific combinations of r-value types. The r-values determined near initial yielding exhibited significant experimental scatter, which resulted in a poor representation of the empirical calibration data. Two calibration methods were evaluated, each of which were dependent on the primary source of empirical data. The accuracy of the initial calibration was evaluated using a numerical model of a hemispherical punch stretch test. The orthotropic sensitivity exhibited by the yield criterion, presented difficulties in obtaining an accurate representation of both the yield behaviour and the strain response. The use of the r-values as the primary source of calibration data resulted in the least accurate description of the anisotropic yield behaviour of the 2024-T3 alloy. The accurate prediction of either the first or second principal strains from the numerical model highlighted the yield criterion deficiencies. The most accurate strain predictions were obtained using the r-values determined from the plastic strain.