Non-linear optimization of a new robotic induction process for local heat treatment using thermal finite element analysis. Issue 5 (May 2016)

• Record Type: Journal Article
• Authors: Gendron, Mathieu; Boudreault, Éric; Hazel, Bruce; Pham, Xuan-Tan; Champliaud, Henri
• Is Part Of: International journal of advanced manufacturing technology. Volume 84:Issue 5/8(2016)
• Page Start: 1013
• Page End: 1029
• Publication Date: 2016-05
• Subjects: Local heat treatment -- Induction heating modeling -- Non-linear optimization -- Finite element analysis -- Robotic process -- Crack repairs -- Cavitation damage repairs -- Erosion damage repairs; Manufacturing processes -- Periodicals; Production engineering -- Periodicals; 670.427
• DOI: 10.1007/s00170-015-7765-z
• Languages: English
• ISSNs: 0268-3768

Abstract

Abstract Performing high-quality repair on aging hydro power equipment is a challenging issue for utilities. Weld repair deteriorates the mechanical properties of the base metal in and around the heat-affected zone. For martensitic stainless steel runners, there is no way to perform post-weld heat treatment (PWHT) on site to restore those properties without dismantling, a very expensive job for such large components, typical of power utilities. To perform in situ high-quality repairs on such components, a new robotic heat treatment process is developed. Heat is generated and controlled using a flat spiral coil mounted on a compact, portable robot and moved over the area needing heat treatment. Unlike conventional induction heating, which requires a customized coil, this new approach combines a universal coil and a flexible robot to heat a broad range of complex shapes. One critical aspect is to set heating and path parameters in order to generate a target spatial and temporal temperature field. This paper proposes a numerical method combining thermal finite element analysis and a non-linear optimization algorithm to set these parameters. The temperature resulting from the electromagnetic field induced by the coil is modeled using an average heat input source to improve computation speed. Good agreement is obtained between numerical and experimental results for PWHT under laboratory conditions.