Scholarlink

<p>The hot deformation behavior of Ferrium® C64 (Fe–16Co–7.5Ni–0.1C) alloy was investigated through isothermal uniaxial hot compression tests over a temperature range of 1123–1423&#xa0;K and strain rates from 0.01 to 10&#xa0;s⁻<sup>1</sup>. Based on the experimental results, three predictive models were developed: Arrhenius-type constitutive model (ACM), strain-compensated Arrhenius-type constitutive model (SACM), and an artificial neural network (ANN) model. Model performance was evaluated using correlation coefficient (R), relative percentage error, and average absolute relative error (AARE). The ANN model demonstrated superior predictive accuracy with an R value of 0.9991 and an AARE of 0.12%, outperforming the ACM (R = 0.9763, AARE = − 8.9%) and SACM (R = 0.9798, AARE = − 9%). At lower strain rate (0.01&#xa0;s<sup>−1</sup>) all models demonstrated good predictability (R &gt; 0.99), but their accuracy decreased at higher strain rates (10&#xa0;s<sup>−1</sup>), with the ANN model showing better performance (R &gt; 0.99) across varying conditions, making it a more reliable tool for predicting flow stress in Ferrium® C64 alloy. Based on the experimental and predicted flow curves, forging simulation for input pinion for military helicopter transmission assembly using DEFORM 3D software was performed and validated through actual forging process.</p> Graphical Abstract <p></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Comparative study on constitutive modeling and artificial neural networks of hot deformation behavior of Ferrium® C64 case carburizing steel and simulation of input pinion for military helicopter transmission assembly

  • 期刊论文

  • 发表日期：

    2025年4月8日

查看全文
  • A. Jayanthi,
  • S. H. Adarsh,
  • Virendra Ahirwar,
  • B. Ravisankar

摘要

The hot deformation behavior of Ferrium® C64 (Fe–16Co–7.5Ni–0.1C) alloy was investigated through isothermal uniaxial hot compression tests over a temperature range of 1123–1423 K and strain rates from 0.01 to 10 s⁻1. Based on the experimental results, three predictive models were developed: Arrhenius-type constitutive model (ACM), strain-compensated Arrhenius-type constitutive model (SACM), and an artificial neural network (ANN) model. Model performance was evaluated using correlation coefficient (R), relative percentage error, and average absolute relative error (AARE). The ANN model demonstrated superior predictive accuracy with an R value of 0.9991 and an AARE of 0.12%, outperforming the ACM (R = 0.9763, AARE = − 8.9%) and SACM (R = 0.9798, AARE = − 9%). At lower strain rate (0.01 s−1) all models demonstrated good predictability (R > 0.99), but their accuracy decreased at higher strain rates (10 s−1), with the ANN model showing better performance (R > 0.99) across varying conditions, making it a more reliable tool for predicting flow stress in Ferrium® C64 alloy. Based on the experimental and predicted flow curves, forging simulation for input pinion for military helicopter transmission assembly using DEFORM 3D software was performed and validated through actual forging process.

Graphical Abstract