Scholarlink

<p>High-temperature shape memory alloys (SMAs) are essential for advanced actuation in aerospace and energy systems, yet their continued development is constrained by challenges in both equilibrium and metastable phase prediction, transformation temperature control, and material sustainability. This study presents a data-driven framework for discovery and multi-objective optimization of novel, concentrated, multi-component SMA compositions that integrates high-throughput CALPHAD (calculation of phase diagrams) with machine learning (ML). The framework is used to identify regions of B2 phase which have potential for exhibiting transitions to the B19’ phase upon cooling. An Extra Trees regression model was trained to predict martensitic and austenitic transformation temperatures. Compositions were selected for experimental validation using cost and supply chain risk metrics to promote practical viability. Selected compositions were synthesized via arc melting for experimental validation using differential scanning calorimetry and dilatometry. The integrated approach enables rapid exploration of complex compositional spaces, linking thermodynamic calculations with functional behavior and real-world constraints. Strengths and limitations of ML in the alloy optimization context are exposed and discussed, along with prospects of such a framework as a scalable pathway for accelerating the discovery and optimization of high-performance SMAs with tailored transformation behavior.</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

Computational Discovery of Medium-Entropy SMAs using CALPHAD, Machine Learning, and Economic Analysis

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月15日

查看全文
  • S. Josyula,
  • A. A. Oladipo,
  • U. P. Ochieze,
  • J. D. Maile,
  • Y. Noiman,
  • N. Simpson,
  • E. J. Payton

摘要

High-temperature shape memory alloys (SMAs) are essential for advanced actuation in aerospace and energy systems, yet their continued development is constrained by challenges in both equilibrium and metastable phase prediction, transformation temperature control, and material sustainability. This study presents a data-driven framework for discovery and multi-objective optimization of novel, concentrated, multi-component SMA compositions that integrates high-throughput CALPHAD (calculation of phase diagrams) with machine learning (ML). The framework is used to identify regions of B2 phase which have potential for exhibiting transitions to the B19’ phase upon cooling. An Extra Trees regression model was trained to predict martensitic and austenitic transformation temperatures. Compositions were selected for experimental validation using cost and supply chain risk metrics to promote practical viability. Selected compositions were synthesized via arc melting for experimental validation using differential scanning calorimetry and dilatometry. The integrated approach enables rapid exploration of complex compositional spaces, linking thermodynamic calculations with functional behavior and real-world constraints. Strengths and limitations of ML in the alloy optimization context are exposed and discussed, along with prospects of such a framework as a scalable pathway for accelerating the discovery and optimization of high-performance SMAs with tailored transformation behavior.