Integrated Taguchi optimization and post-weld heat treatment effects on microstructural evolution and mechanical performance of friction stir welded AA3103 aluminium alloy
摘要
The present study aims to optimize friction stir welding parameters for AA3103 using a Taguchi L9 orthogonal array and to evaluate the mechanical consistency of post-weld heat treatment (PWHT) joints through signal-to-noise (S/N) ratio analysis, ANOVA, and regression modeling. The investigated parameters include tool pin profile (square, circular, triangular), shoulder-to-pin diameter ratio (D/d = 3, 4, 5), rotational speed (355, 710, 1400 RPM), and welding speed (20, 40, 80 mm/min). After welding, joints were subjected to a controlled PWHT cycle at 350 °C for 2 h followed by artificial aging at 200–225 °C and water quenching. Mechanical performance was assessed in terms of ultimate tensile strength (UTS), yield stress (YS), and percentage elongation in accordance with ASTM E8 standards. Results demonstrate that the shoulder-to-pin ratio remains the dominant factor governing strength retention after PWHT (delta = 0.99 for UTS, delta = 1.57 for YS), whereas rotational speed significantly influences ductility stability (delta = 0.41 for elongation). The optimal parameter combination was identified as a circular pin profile, D/d = 3, rotational speed of 1400 RPM, and welding speed of 40 mm/min. PWHT substantially reduced mechanical scatter; furthermore, regression prediction accuracy was markedly improved after PWHT, with most prediction errors confined within ± 7–8%, demonstrating the statistical benefit of microstructural stabilization on model reliability. Microstructural analysis confirmed fine equiaxed grain retention in the stir zone, reduced dislocation density, smoother grain transitions, and minimized hardness gradients after PWHT. The integrated FSW–PWHT optimization framework established in this work provides a statistically robust and industrially relevant methodology for achieving repeatable mechanical performance in AA3103 joints.
Graphical abstract