Prediction of the keyhole TIG welding-induced distortions on Inconel 718 industrial gas turbine component by numerical-experimental approach

Abstract Welding technologies represent a paramount joining process for ensuring the quality and reliability of critical industrial components; therefore, their innovation constitutes a driving force in realizing increasingly competitive products. A recently developed technology is the keyhole TIG welding, a new high energy–density alternative to the conventional TIG process. A key role in improving innovative manufacturing processes such as the keyhole TIG is covered by numerical simulation; indeed, it allows the development of a process digital twin able to support decisions and work as a predictive tool. Within this framework, the paper deals with the numerical-experimental investigation of the keyhole TIG technology, successfully employed on a simplified mock-up of an industrial gas turbine component consisting of two 6.5-mm-thick Inconel 718 rings. Numerical analysis aimed at predicting welding-induced distortions was performed employing two different computational approaches, namely the moving heat source and the simplified imposed thermal cycle methods. The numerical-experimental comparison of the results demonstrates an innovative approach in the field of the current keyhole TIG numerical simulation since, besides verification of numerical thermal analysis, further substantial validation of the post-weld distortion predictions is provided through comprehensive three-dimensional experimental data. Moreover, the comparative assessment of the two computational approaches and experimental evidence revealed that the imposed thermal cycle method implementation does not compromise the accuracy of welding distortion forecasting in industrial applications such as that investigated. Therefore, it can be regarded as a valuable tool for supporting the process engineer in designing the ideal set-up to comply with a variety of industrial requirements, among them strict design tolerances.