Repetitive upsetting extrusion (RUE) was applied to the as-homogenized Mg-12.0Gd-4.5Y-2Zn-0.4Zr (wt %) alloy at 773 K. The microstructure evolution of the alloy during RUE was investigated. The results indicated that almost all Mg5(Gd,Y,Zn) phases and fine-lamellar long-period stacking-ordered (LPSO) phases were dissolved into the matrix after homogenization treatment at 793 K for 16 h. After one RUE pass, dynamic recrystallization (DRX) occurred. During subsequent RUE passes (from one to three passes), average volume fractions of DRXed grains increased from 43.9% to 65.8%, and that of fine-lamellar and block-shaped LPSO phases gradually decreased. All samples exhibited a typical bimodal microstructure consisting of some initial grains containing fine-lamellar LPSO phases, but consisting mostly of fine-DRXed grains with a mean grain size of 6 μm. Because of an increase in the accumulated strains, the coarse grains were substituted with fine-DRXed grains, the grains were gradually refined, and the microstructure distribution became more homogeneous.

The decreasing temperature reciprocating upsetting-extrusion (RUE) deformation experiment was carried out on Mg-Gd-Y-Zr alloy to study RUE deformation on the influence of microstructure of the alloy. This work showed that with the gradual increase of RUE deformation passes, the continuous dynamic recrystallization (CDRX) process and the discontinuous dynamic recrystallization (DDRX) process occurred at the same time, and the grain refinement effect was obvious. Particulate precipitation induced the generation of DRX through particle-stimulated nucleation (PSN). In addition, after one pass of RUE deformation, the alloy produced a strong basal texture. As the RUE experiment proceeded, the basal texture intensity decreased. The weakening of the texture was due to the combined effect of DRX and alternating loading forces in the axial and radial directions. After four RUE passes, the mechanical properties of the alloy had been significantly improved, which was the result of the combined effect of dislocation strengthening, fine grain strengthening, and second phase strengthening.

The effects of repetitive upsetting extrusion under different starting temperatures on the microstructure and texture of GWZK124 alloy were investigated. The results clearly showed that the particles and second phases induced dynamic recrystallization (DRX), which can be explained by the particle-stimulated nucleation (PSN) mechanism. It was shown that grain refinement during repetitive upsetting extrusion (RUE) is dominated by a complicated combination of continuous dynamic recrystallization and discontinuous dynamic recrystallization. The RUEed alloys under different starting temperatures exhibited a bimodal microstructure comprising fine DRXed grains with weak texture and coarse deformed grains with strong texture. The DRXed grains could weaken the texture. As the RUE starting temperature decreased, the average grain size increased and the volume fraction of DRXed grains decreased.

The effect of friction coefficient, reduction of thickness, forming speed, roller diameter on the coefficient of forward slip at the rim of the roller during cup roll forming was studied by numerical simulation. The results show that during roll pre-forming, the friction coefficient and forming speed have little effect on the coefficient of forward slip, and the reduction of thickness and the diameter of the roller have a greater influence on the coefficient of forward slip; as the reduction of thickness increases, the coefficient of forward slip increases; as the roller diameter increases, the coefficient of forward slip increases first and then decreases. The coefficient of forward slip during the roll pre-forming of cup is much larger than that of roll forging. During roll shaping, the process parameters have little effect on the coefficient of forward slip and its value is close to zero, that is, the axial length of the roll shaped part is almost equal to the arc length of the roller rim. The accuracy of numerical simulation is verified by experiments, and the error between the numerical simulation and experiment is less than 1%.