Abstract: Although the {10-12} twinning behavior of Mg alloys under uniaxial tension and compression has been extensively investigated, the simulations of {10-12} twinning behavior under biaxial tension have rarely been reported. In this work, the EVPSC-TDT model is first employed to systematically investigate the deformation behavior of a Mg alloy AZ31 plate under biaxial tension in the RD-TD and ND-TD planes. The RD, TD and ND refer to the rolling direction, transverse direction, and normal direction of the hot rolled plate. The measured stress-strain curves and texture evolutions are well predicted and the contours of plastic work under biaxial tension are also constructed for comparison with experiments. The plastic response has been interpreted in terms of relative activities of various deformation modes. For biaxial tension in the RD-TD plane, basal and pyramidal slips mainly contribute to the plastic deformation for stress ratios of ${{\sigma }_{\text{RD}}}:{{\sigma }_{\text{TD}}}=$1:2 to 2:1. Prismatic slip becomes more active for ${{\sigma }_{\text{RD}}}:{{\sigma }_{\text{TD}}}=1:4$ and 4:1. Compression twinning could be activated and so cause texture reorientation at large strains, especially for ${{\sigma }_{\text{RD}}}:{{\sigma }_{\text{TD}}}=1:1$. The six-fold feature of {10-10} pole figure could still be observed for ${{\sigma }_{\text{RD}}}:{{\sigma }_{\text{TD}}}=1:4$ and 4:1 at large strain. For biaxial tension in the ND-TD plane, tensile twinning plays an important role for ${{\sigma }_{\text{ND}}}:{{\sigma }_{\text{TD}}}\ge 1:2$, while prismatic slip contributes to plastic deformation for the other cases. With the increase of stress ratio from ${{\sigma }_{\text{ND}}}:{{\sigma }_{\text{TD}}}\ge 1:1$ to 1:0, the predicted twin volume fractions (VFs) at a specific strain along the ND, ${{\varepsilon }_{\text{ND}}}$, almost linearly decrease, however, it is seen that the experimental ones at given strains along the ND do not follow such a trend with the measured twin VFs within the range of stress ratios, $2:1\le {{\sigma }_{\text{ND}}}:{{\sigma }_{\text{TD}}}\le 6:1$, clearly being overestimated, and the difference between experiments and simulations becomes most obvious at the relatively small strain of ${{\varepsilon }_{\text{ND}}}=0.015$. The possible reasons for the observed difference are discussed.

Key words: Magnesium alloy, Biaxial tension, Simulation, Twinning