Key Lab of Electromagnetic Processing of Materials, Ministry of Education, Northeastern University, Shenyang, China and College of Materials Science and Engineering, Northeastern University, Shenyang, China
Auto-generated profile, based on publicly available information
Wenchao Duan
Key Lab of Electromagnetic Processing of Materials, Ministry of Education, Northeastern University, Shenyang, China and College of Materials Science and Engineering, Northeastern University, Shenyang, China
Purpose: The purpose of this paper is to investigate the effect of pulsed magnetic field (PMF) with different duty cycles on the melt flow and heat transfer behaviors during direct-chill (DC) casting of large-size magnesium alloy billet and find the appropriate range of duty cycle. Design/methodology/approach: A transient two-dimensional mathematical model coupled electromagnetic field, flow field and thermal field, is conducted to study the melt flow and temperature field under PMF and compared with that under the harmonic magnetic field. Findings: The results reveal that melt vibration and fluctuation are generated due to the instantaneous impact of repeated thrust and pull effects of Lorentz force under PMF. The peak of Lorentz force decreases greatly with the increasing duty cycle, but the melt fluctuation region is expanded with higher duty cycle, which accelerates the interior melt velocity and reduces the temperature gradient at the liquid-solid interface. However, PMF with overly high duty cycle has adverse effect on the melt convection and limited influence on the interior melt. A duty cycle of 20% to 50% is a reasonable range. Practical implications: This paper can provide guiding significance for the setting of duty cycle parameters on DC casting under PMF. Originality/value: There are few reports on the effect of PMF parameters during DC casting with applying PMF, especially for duty cycle, a parameter unique to PMF. The findings will be helpful for applying the external field of PMF on DC casting.
Accurate prediction of hot-deformation behaviour in low-cost armour-grade titanium alloys is difficult because phenomenological equations incompletely represent temperature-strain-rate coupling, whereas purely data-driven models provide limited physical interpretability. Here, a K-Arrhenius physics-informed neural network (K-A-PINN) is developed for TC4A alloy using 48 corrected flow curves acquired at 750–1025 °C and 0.01–10 s−1. The framework combines a strain-compensated Arrhenius baseline, a K-factor correction and a gated affine mapping that learns condition-dependent constitutive parameters subject to physical bounds. The K-factor correction increased R2 from 0.9500 to 0.9939 and reduced the RMSE from 25.62 to 8.97 MPa. Under withheld deformation conditions, K-A-PINN achieved R2 = 0.9954, RMSE = 4.7954 MPa and MAPE = 3.2115%. Analysis of the learned variables separated five constitutive-response regimes. Processing maps identified a high-efficiency domain at 900–950 °C and 0.01–0.1 s−1. EBSD observations showed incomplete boundary-associated recrystallization at 850 °C and a strain rate of 0.01 s−1, recovery-dominated softening at 925 °C and a strain rate of 0.01 s−1 and substantial retained deformation at 925 °C and a strain rate of 10 s−1. The model therefore links flow-stress prediction to experimentally observed changes in microstructural response within the investigated processing window.