The purpose of the present study is to describe and characterize the anisotropic flow and fracture behaviour of a high-strength aluminium alloy. To this end, thick plates of AA7075-T651 aluminium alloy have been tested. Different specimen geometries were used to investigate various stress states. Each specimen was machined in different directions of the plate to enlighten the anisotropy of the material. For all tests, the plastic flow exhibited a slight anisotropy while the failure strain and failure modes showed a very important dependence to the loading direction.
A microstructure analysis of the virgin material was performed by scanning electron microscope (SEM) and electron back-scatter diffraction to identify its texture, grain shape and particle distribution. A transmission electronic microscope analysis gave information of the precipitate free zones and their composition.
Tensile tests were performed on smooth axisymmetric specimens under uniaxial tension. Tensile tests were also conducted on notched axisymmetric specimens of notch radii and to obtain higher stress triaxiality states. Shear tests were performed on butterfly specimens and compression tests were performed on cylindrical specimens. Fracture surface analyses were carried out by SEM to identify the failure modes, supported by the microstructure analysis.
Based on the plastic anisotropy observed experimentally, the Yld2004-18p anisotropic yield function proposed by Barlat et al. (2005) was chosen to model the elasto-plastic behaviour of the AA7075-T651 alloy. The plastic parameters were calibrated using seven in-plane uniaxial tensile tests, a compression test in the normal direction of the plate and a shear test in the rolling direction. Numerical simulations of all the experimental tests were performed using the anisotropic elasto-plastic model. Predicted stress-strain curves were in very good agreement with the experimental curves for all tests including the tensile tests on notched specimens, which were not used in the calibration of the model. The overestimation of predicted stress level, generally observed (e.g. by Wilson, 2002) with notched specimens and isotropic pressure independent yield function, was significantly decreased when taking into account anisotropy. The stress and strain states on elements where failure is experimentally observed were evaluated. The establishment of a failure locus (relation between failure strain and stress triaxiality) was also discussed.
Analytical approaches were used to gain some insight of the failure process. First, the void growth approach proposed by Rice and Tracey (1969) was extended to an anisotropic matrix. Then, the usual localization criterion (Rice, 1976) was developed with various constitutive characteristics to account for the shape of the yield function, non-associative plastic flow, large deformations and thermo-mechanical couplings.
For industrial applications, a phenomenological failure criterion based on “plastic work”, called the anisotropic extended Cockcroft-Latham (AECL), was proposed. The criterion was calibrated using the seven uniaxial in-plane tensile tests and the shear test performed in the rolling direction. Numerical simulations of all tests were, once again, performed accounting for plastic anisotropy. A parameter study was carried out to enlighten the influence of parameters such as the plastic anisotropy and the failure anisotropy. The predicted failure strain and failure modes were not accurate enough to give predictive capability to this failure criterion in all material tests.
Finally, this anisotropic failure criterion was also used in numerical simulations of some impact tests on AA7075-T651 plates with ogival and blunt projectiles. A thermoelasto-thermoviscoplastic model with anisotropic yielding was used and as for the material tests, a parameter study was performed. Ballistic limits were predicted and compared with the experimental results obtained by Børvik et al. (2010). It was found that the anisotropy of plastic flow and failure had almost no influence at very high impact velocities, while it had a substantial effect at impact velocities close to the ballistic limit. The introduced anisotropy was not found to improve the ballistic limit prediction for all cases, and also other parameters (e.g. yield shape, temperature coefficients and contact algorithms) have a prominent influence on the predicted ballistic limit. However, supported by experimental observations of non-axisymmetric failure modes (Pedersen et al., 2011), both the plastic anisotropy and the failure anisotropy are believed to be important ingredient of the constitutive model.