Scholarlink

<p>Conventional phase-field approaches employed in existing studies fail to quantitatively account for dislocation-interface interactions, resulting in a predictive error exceeding 35% in degradation rate. Furthermore, the absence of competitive mechanisms among metastable phases in CALPHAD (Calculation of Phase Diagrams) thermodynamic databases imposes limitations on Molecular Dynamics (MD) simulations, particularly with respect to temporal scales, thereby hindering accurate modeling of hour-long thermal exposure processes. This study introduces a coupled Phase-Field Crystal-CALPHAD (PFC-CALPHAD) framework integrated with a Convolutional Neural Network (CNN)-based nucleation model dependent on dislocation density, along with a dynamic database-updating Fast Fourier Transform (FFT) algorithm. This integrated approach establishes a simulation system to predict microstructural degradation of Bainite in fire-resistant steel under 600°C/4&#xa0;h thermal exposure, reducing the prediction error of reversed austenite content to 1.8%. Experimental validation via Electron Backscatter Diffraction (EBSD) was employed to track the migration of bainite/austenite phase boundaries within a coupled temperature-time field. The results verify that solute segregation, driven by carbon activity gradients at the bainite/austenite interface, serves as the predominant factor governing degradation. This finding provides a new theoretical basis for assessing the high-temperature service life of fire-resistant steels.</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

Bainite Degradation in Simulated Fire-Resistant Steel Exposed at 600°C for 4 Hours by Coupling Phase-Field Crystal Method and CALPHAD

  • 期刊论文

  • 发表日期：

    2026年3月5日

查看全文
  • Liang Song,
  • Yun Peng,
  • Haiyan Zhao,
  • Lin Zhao

摘要

Conventional phase-field approaches employed in existing studies fail to quantitatively account for dislocation-interface interactions, resulting in a predictive error exceeding 35% in degradation rate. Furthermore, the absence of competitive mechanisms among metastable phases in CALPHAD (Calculation of Phase Diagrams) thermodynamic databases imposes limitations on Molecular Dynamics (MD) simulations, particularly with respect to temporal scales, thereby hindering accurate modeling of hour-long thermal exposure processes. This study introduces a coupled Phase-Field Crystal-CALPHAD (PFC-CALPHAD) framework integrated with a Convolutional Neural Network (CNN)-based nucleation model dependent on dislocation density, along with a dynamic database-updating Fast Fourier Transform (FFT) algorithm. This integrated approach establishes a simulation system to predict microstructural degradation of Bainite in fire-resistant steel under 600°C/4 h thermal exposure, reducing the prediction error of reversed austenite content to 1.8%. Experimental validation via Electron Backscatter Diffraction (EBSD) was employed to track the migration of bainite/austenite phase boundaries within a coupled temperature-time field. The results verify that solute segregation, driven by carbon activity gradients at the bainite/austenite interface, serves as the predominant factor governing degradation. This finding provides a new theoretical basis for assessing the high-temperature service life of fire-resistant steels.