Representing the Effect of Crystallographic Texture on the Anisotropic Performance Behavior of Rolled Aluminum Plate

Abstract

Rolled aluminum alloys are known to be anisotropic due to their processing histories. This paper focuses on measuring and modeling monotonic and cyclic strength anisotropies as well as the associated anisotropy of the elastic/elastic-plastic transition of a commercially-available rolled plate product. Monotonic tension tests were conducted on specimens in the rolling plane of 25.4 mm thick AA 7075-T6 plate taken at various angles to the rolling direction (RD). Fully-reversed tension/compression cyclic experiments were also conducted. As expected, we found significant anisotropy in the back-extrapolated yield strength. We also found that the character of the elastic/elastic-plastic transition (knee of the curve) to be dependent on the orientation of the loading axis. The tests performed in RD and TD (transverse direction) had relatively sharp transitions compared to the test data from other orientations. We found the cyclic response of the material to reflect the monotonic anisotropy. The material response reached cyclic stability in 10 cycles or less with very little cyclic hardening or softening observed. For this reason, we focused our modeling effort on predicting the monotonic response. Reckoning that the primary source of anisotropy in the rolled plate is the processing-induced crystallographic texture, we employed the experimentally-measured texture of the undeformed plate material in continuum slip polycrystal plasticity model simulations of the monotonic experiments. Three types of simulations were conducted, upper and lower bound analyses and a finite element calculation that associates an element with each crystal in the aggregate. We found that all three analyses predicted anisotropy of the back-extrapolated yield strength and post-yield behavior with varying degrees of success in correlating the experimental data. In general, the upper and lower bound models predicted larger and smaller differences in the back-extrapolated yield strength, respectively, than was observed in the data. The finite element results resembled those of the upper bound when initially cubic elements were employed. We found that by employing an element shape that was more consistent with typical rolling microstructure, we were able to improve the finite element prediction significantly. The anisotropy of the elastic/elastic-plastic transition predicted by each model was also different in character. The lower bound predicted sharper transitions than the upper bound model, capturing the shape of the knee for the RD and TD data but failing to capture the other orientations. In contrast, the upper bound model predicted relatively long transitions for all orientations. As with the upper bound, the FEM calculation predicted gentle transitions with less transition anisotropy predicted than that of the upper bound. [S0094-4289(00)00201-2]