材料科学
焊接
高温合金
粒度
振幅
线性回归
摩擦焊接
振荡(细胞信号)
有限元法
回归分析
冶金
机械
热力学
数学
统计
微观结构
物理
生物
遗传学
量子力学
作者
Mahshad Javidikia,Morteza Sadeghifar,Henri Champliaud,Mohammad Jahazi
标识
DOI:10.1016/j.jajp.2023.100150
摘要
This research study was aimed at investigating the influence of linear friction welding parameters on grain size
\nalteration and temperature distribution of Ni-base superalloy Waspaloy. A 3D finite element model was developed
\nto predict average grain size and peak temperature as responses. The linear friction welding parameters
\nconsisted of oscillation amplitude, oscillation frequency, and applied pressure. Initially, the evolution of the
\naverage grain size as a function of the most influential process parameters was subsequently modeled based on
\nthe Johnson-Mehl-Avrami-Kolmogorov recrystallization model and were then validated with experimental results.
\nThen, D-optimal design of experiments and analysis of variance were conducted to determine the most
\ninfluential process parameters that affect the average grain size and peak temperature of the welded joint.
\nThereafter, response surface method was employed to obtain the regression models of the responses. The analysis
\nof variance demonstrated that the P-value of the regression models was smaller than 5% and R2, R2
\nadj, and R2
\nPred
\nwere between 87% and 97%, which showed that the predictive regression models of PT and AGS can be used
\nwith a high level of confidence. The regression models were then validated by selecting two extra LFW tests in
\nthe space of the DoE. The optimum values of the welding parameters were determined to minimize the responses.
\nThe multi-criteria optimization analysis showed that both average grain size and peak temperature were more
\ndependent on pressure than oscillation amplitude and frequency. The developed finite element and regression
\nmodels can be utilized as a predictive tool for the design of joining industrial components, which minimize
\nexpensive and time-consuming experimental tests and measurements.
科研通智能强力驱动
Strongly Powered by AbleSci AI