Neural network-based hardness and toughness prediction in HAZ of temper bead welding repair technology

• Published in: QUARTERLY JOURNAL OF THE JAPAN WELDING SOCIETY Vol. 31; no. 4; pp. 173s - 177s
• Main Authors: Yu, Lina, Sasa, Masato, Ohnishi, Kenta, Kameyama, Masashi, Hirano, Shinro, Chigusa, Naoki, Sera, Takehiko, Saida, Kazuyoshi, Mochizuki, Masahito, Nishimoto, Kazutoshi
• Format: Journal Article
• Language: English
• Published: JAPAN WELDING SOCIETY 2013
• Subjects: FEM; Hardness; low-alloy steel; Temper bead welding; Toughness
• ISSN: 0288-4771; 2434-8252
• DOI: 10.2207/qjjws.31.173s

The aged nuclear power plant needs to be repaired or maintained, and temper bead welding is one effective repair welding methods instead of post weld heat treatment. For temper bead welding, hardness and toughness are the key criteria to evaluate the tempering effect. A neural network-based method for hardness and toughness prediction in heat affected zone of low-alloy steel has been investigated to evaluate the tempering effect in temper bead welding. On the basis of experimentally obtained database, the new hardness and toughness prediction system was constructed by using RBF-neural network. With it, the hardness and toughness distribution in heat affected zone of temper bead welding was calculated based on the thermal cycles numerically obtained by Finite Element Method. The predicted hardness and toughness were in good accordance with the experimental results. It follows that our new prediction system is effective for estimating the tempering effect during temper bead welding and hence enables us to assess the effectiveness of temper bead welding before the actual repair welding.