Scholarlink

<p>As oil and gas exploration expands to deep-sea, ultra-deep, and unconventional reservoirs, demands for nickel-based corrosion-resistant alloys' mechanical performance grow stricter. Traditional trial-and-error alloy development is time-consuming and costly, making them inadequate for rapid alloy design. This study developed an interpretable composition–microstructure–property optimization model by integrating genetic algorithm, machine learning, and thermodynamic calculations to enable fast design of ultra-high-strength age-hardened nickel-based corrosion-resistant alloys. Thermo-Calc (via TC-Python) calculated γ′/γ″ phase volume fractions and precipitation driving forces under different compositions, used as input for a machine learning-based yield strength prediction model. An AdaBoost regressor was trained and embedded into the genetic algorithm as the fitness function to perform constrained composition optimization. The optimized alloy, ML1, exhibited excellent mechanical performance under various heat treatment conditions. Notably, after solution treatment at 1030 °C followed by aging, it achieved a yield strength of 1365 MPa and an ultimate tensile strength of 1539 MPa—significantly outperforming most commercial precipitation-hardened nickel-based corrosion-resistant alloys. The proposed optimization framework effectively reduces development time and cost compared to conventional methods and is extendable to other precipitation-hardened alloy systems. It offers a new strategy for data-driven alloy design with both scientific value and engineering applicability.</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

Genetic algorithm coupled with thermodynamic calculations for high-strength nickel-based corrosion-resistant alloys design

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月13日

查看全文
  • Yingjie Sun,
  • Yong Lian,
  • Rongrong Chen,
  • Piao Qian,
  • Jin Zhang,
  • Yinghu Wang,
  • Qubo He,
  • Dadi Zhou,
  • Hengcan Yang

摘要

As oil and gas exploration expands to deep-sea, ultra-deep, and unconventional reservoirs, demands for nickel-based corrosion-resistant alloys' mechanical performance grow stricter. Traditional trial-and-error alloy development is time-consuming and costly, making them inadequate for rapid alloy design. This study developed an interpretable composition–microstructure–property optimization model by integrating genetic algorithm, machine learning, and thermodynamic calculations to enable fast design of ultra-high-strength age-hardened nickel-based corrosion-resistant alloys. Thermo-Calc (via TC-Python) calculated γ′/γ″ phase volume fractions and precipitation driving forces under different compositions, used as input for a machine learning-based yield strength prediction model. An AdaBoost regressor was trained and embedded into the genetic algorithm as the fitness function to perform constrained composition optimization. The optimized alloy, ML1, exhibited excellent mechanical performance under various heat treatment conditions. Notably, after solution treatment at 1030 °C followed by aging, it achieved a yield strength of 1365 MPa and an ultimate tensile strength of 1539 MPa—significantly outperforming most commercial precipitation-hardened nickel-based corrosion-resistant alloys. The proposed optimization framework effectively reduces development time and cost compared to conventional methods and is extendable to other precipitation-hardened alloy systems. It offers a new strategy for data-driven alloy design with both scientific value and engineering applicability.