A Preliminary Study on the CLAM Steel Composition Optimization Based on Extreme Learning Machine

• Published in: Journal of fusion energy 2015-10, Vol.34 (5), p.1071-1076
• Main Authors: Shen, Longfeng, Zhai, Xiangwei, Chen, Chaobin, Li, Chunjing, Wang, Fang
• Format: Article
• Language: English
• Subjects: Activation; Artificial neural networks; Blankets (fusion reactors); Breeder reactors; China; Composition; Dual phase steels; Energy Systems; Engineering test reactors; Ferritic stainless steels; Fusion reactors; Grain size; Learning; Machine learning; Martensitic stainless steels; Mechanical properties; Metal fatigue; Neural networks; Neurons; Nuclear Energy; Nuclear Fusion; Nuclear power generation; Nuclear power plants; Optimization; Original Research; Physics; Physics and Astronomy; Plasma Physics; Steel; Structural steels; Sustainable Development; Tensile strength; Yield stress
• ISSN: 0164-0313; 1572-9591
• DOI: 10.1007/s10894-015-9912-9

Material problem is one of the key issues for the realization of fusion energy. Reduced Activation Ferritic/Martensitic steels have been considered to be the primary candidate structural material for fusion DEMO reactors and the first fusion power plants. China low activation martensitic steel (CLAM) has been chosen as the primary structure material in the designs of FDS series PbLi blankets for fusion reactors, CN helium cooled ceramic breeder test blanket module for ITER, liquid blanket of China fusion engineering test reactor (CFETR). In this work, optimization of CLAM steel composition was performed based on extreme learning machine method. The results showed that CLAM steel with Ta content of 0.18–0.20 % had better tensile property at the temperature of 350–550 °C, which provided references for the optimization of CLAM steel composition.