Tube roller compression (TRC): A novel severe plastic deformation technique for grain refinement and enhanced mechanical properties of tubular materials

Abstract

• Significant Grain Refinement and Microstructure Improvement : The TRC technique achieved up to 68% grain size reduction in AA6014 alloy, with improved microstructure uniformity and higher LAGB density, particularly at lower rotational speeds. • Enhanced Mechanical Properties : TRC processing resulted in a 70.5% increase in surface hardness , as well as improvements in ultimate tensile strength (48%) and yield strength (53%) . • Effective Prediction and Mechanisms : Machine learning models showed 95% accuracy in predicting experimental outcomes, with enhancements in hardness and strength attributed to strain hardening, grain refinement, and increased dislocation density, although ductility decreased. A novel severe plastic deformation (SPD) method named Tube Roller Compression (TRC) is introduced for processing aluminum alloy tubes with scalable dimensions. The process uses a rubber pad to generate back pressure as a rotating roller tool compresses a constrained tube, locally deforming and sequentially refining its structure. Experiments on AA6014 tubes showed a significant grain size reduction of 66% after 3 cycles compared to the as-annealed counterpart, with notable improvements in yield and ultimate strength of 47% and 44.5%, respectively. In terms of surface hardness, TRC through 3 cycles revealed a notable increase in the surface hardness by 70.5% compared to the as-annealed counterpart. Electron backscatter diffraction analysis revealed progressive grain refinement and crystallographic texture. At 45 rpm, grain size reduced from 90.54 µm (as-annealed) to 29.53 µm after 3 cycles, with increased low-angle grain boundaries and dominant {101} orientation. At 180 rpm, grain size reached 38.67 µm after 3 cycles, with a shift toward strong {111} texture. Higher speeds favored refinement but resulted in less uniform microstructures. Statistical analysis via ANOVA and response surface methodologies confirmed the influence of processing parameters on mechanical properties. Additionally, machine learning predicted outcomes with up to 95% accuracy, highlighting TRC’s potential for optimized, data-driven SPD processing.