Developing a realistic FE analysis method for the welding of a NET single-bead-on-plate test specimen

A Net (A European network) single-bead-on-plate test specimen was modeled using FE method (Abaqus) in order to predict the final residual stress in the plate following welding to compare with measured distributions. The welding of the specimen was modeled using 3D uncoupled analysis. Because the complexity of the 3D analysis, sensitivity tests on it were expensive. A 2D FE mesh for the test specimen was set up in order to run these sensitivity tests. The effects of changing heat flux distribution model, choosing different boundary conditions, removing latent heat. choosing different work hardening parameters and changing anneal temperature were tested on the mesh. The results yielded were compared with experimental results and it was found that ellipsoidal power density distribution is suitable to be used in 3D analysis. Fully constraining the edge, which was in contact with a vice was found to be inappropriate as it could cause the underestimating of the predicted results. Changing the annealing temperature or removing latent heat when analyzing the specimen in 3D model were not expected to have obvious effects on the predicted results of the analysis according to the sensitivity tests, while changing the hardening parameter from isotropic to kinematic decreased the predicted residual stresses. Based on the sensitivity tests, two 3D FE analyses using two moving heat source techniques were set up. The thermal analyses were accomplished and the results for both techniques show good agreement with experimental measurements, but the analysis using element birth techniques, which closely simulated the welding procedure, was more accurate. The predictions of mechanical analyses were also verified using experimental results and methods. which could help to get more precise predictions were discussed. (C) 2007 Elsevier B.V. All rights reserved.