Prediction of residual stresses in girth welded pipes using an artificial neural network approach. (February 2017)

• Record Type: Journal Article
• Authors: Mathew, J.; Moat, R.J.; Paddea, S.; Fitzpatrick, M.E.; Bouchard, P.J.
• Is Part Of: International journal of pressure vessels and piping. Volume 150(2017)
• Page Start: 89
• Page End: 95
• Publication Date: 2017-02
• Subjects: Residual stress profile -- Girth welds -- Stainless steel -- Neural network -- Neutron diffraction -- Contour method; Pressure vessels -- Periodicals; Pipe -- Periodicals; Récipients sous pression -- Périodiques; Tuyaux -- Périodiques; Pipe; Pressure vessels; Periodicals; 681.76041
• DOI: 10.1016/j.ijpvp.2017.01.002
• Languages: English
• ISSNs: 0308-0161

Abstract

Abstract: Management of operating nuclear power plants greatly relies on structural integrity assessments for safety critical pressure vessels and piping components. In the present work, residual stress profiles of girth welded austenitic stainless steel pipes are characterised using an artificial neural network approach. The network has been trained using residual stress data acquired from experimental measurements found in literature. The neural network predictions are validated using experimental measurements undertaken using neutron diffraction and the contour method. The approach can be used to predict through-wall distribution of residual stresses over a wide range of pipe geometries and welding parameters thereby finding potential applications in structural integrity assessment of austenitic stainless steel girth welds. Highlights: We model residual stresses in multi-pass girth welds using artificial neural networks. We validated the model with experimental measurements using neutron diffraction and contour method. A histogram network was developed to provide a reliable prediction interval of the estimated stress distributions. The model can function providing the weldment type lie within the boundary of the training data envelope used. ANN model can find potential applications in the structural integrity assessment of weldments.