Papers by Author: Dimitrios Elias Katsareas

Abstract: Finite element prediction of residual stresses in a 3-bead letterbox-type repair weld is investigated in the present study. The repair is performed on a 2¼CrMo low alloy ferritic steel plate, containing a machined central groove where three weld beads are deposited using AL CROMO S 225 2¼CrMo electrodes. The proposed simulation procedure, which is based on decoupled thermal and mechanical analyses and the “birth and death of elements” technique, is evaluated through comparison of predicted stresses with neutron diffraction testing data. Parametric studies include modelling aspects such as 2-D plane strain versus 3-D analysis, re-melting of weld material during sequential bead deposition, melting of base plate near the fusion line and annealing. It is concluded that numerical results come, in general, in satisfied agreement with the experimental data.

Abstract: Dissimilar metal welds are commonly found in the primary piping of pressurized water nuclear reactor power plants. The safety assessment practice for such welds requires residual stresses to be taken into consideration. In the present paper the finite element method is utilized for the simulation of the welding process and prediction of the residual stress field in a dissimilar metal weld pipe joint. Although it is common practice to develop in-house finite element codes for weld simulation, the ANSYS commercial finite element code is selected. This is mainly due to the fact that industry focuses on commercial software, since residual stress analysis procedures based on them can be readily transferred to industrial applications. A simplified 2-D axi-symmetric model, in which residual stresses are produced due to the thermo-mechanical properties mismatch during cooling of the weld, is compared with a detailed model in which the complete multi-pass welding procedure is simulated. The latter incorporates the “birth & death of elements” technique, temperature dependant material properties and kinematic hardening material behavior. The aim of this comparison is to establish the degree of model detail and complexity, necessary to obtain satisfactory results and consequently to define a golden rule between computational cost and practically accurate predictions. Identifying the specific simulation parameters and variables, that have the highest impact on the accuracy of the computed results, is also important. It is concluded that, a bead-by-bead or lump-by-lump detailed simulation is necessary in order to obtain reasonably accurate residual stresses that cannot be predicted by a simplified model. A general conclusion is that the proposed method, being simple in implementation and cost effective concerning model complexity and analysis time, is a potential weld residual stress prediction tool.