Numerical simulation of welding distortion in thin plates

Abstract

A three-dimensional model based on the finite-element method is developed to simulate the temperature field and stress distribution in the welding and heat-affected zones during fusion welding of thin plates. The governing equations are solved using the SYSWELD program commercial code. The model's predic- tions are tested and verified against the experiments. Angular distortion and longitudinal bending are meas- ured, the results are compared with those obtained from the mathematical model, and a relatively good agreement between them is found. The verified model is used to evaluate the effects of various parameters on the temperature and stress distributions in the welding and heat-affected zones of a thin austenitic stain- less steel plate. Introduction. Welding distortion and residual stress have remained major challenges in manufacturing. In the automotive industry, parts involving thin plates are commonly employed, and a lot of such parts are assembled by the arc welding process because of its high productivity. Due to the relatively small stiffness of the plates, significant welding distortion often occurs. To comprehensively understand the characteristics of welding deformation in a struc- ture composed of welded thin plates, it is necessary to carry out further fundamental investigations by means of both experimentation and numerical simulation. Extensive experimental efforts have been made to understand and control the welding distortions and residual stresses. In the last two decades, numerous results on using analytical and computational methods have also been re- ported to predict and quantify the effect of the distortion and residual stresses during the welding process. However, there is a very limited literature concerning the prediction and measurement of welding deformation in structures in- volving welded thin plates, especially for welded structures with plate or wall thickness below 3 mm. For a given con- figuration, the magnitude of those effects depends primarily on the specific thermal energy input of the welding process, that is, on the energy per unit length of weld. Therefore, significant reductions in distortion cannot be achieved without using the welding process with a smaller heat input, which is not always practicable. An accurate prediction of distortion is difficult even for simple welded structures, and all the more for fabrications where several such processes are applied consecutively. In previous works (1, 2), simulation of the welding process for a plate from CT3 steel with a thickness of 3 mm and a plate from austenitic stainless steel with a thickness of 1 mm was carried out. In (2), due to the restriction of the Ls-Dyna program in the removal of material history at high temperatures, the behavior of steel during welding was assumed to be fully plastic. However, many authors suggest that the behavior of metals during welding obeys the model of isotropic or kinematic hardening. Because of this, in the present work the isotropic hardening model is con- sidered. The simulation is accomplished by the SYSWELD program, which is a famous software in simulation of welding, and all the physical and mechanical properties of steel are taken from its library. As a first step towards simplification, the sequential computations for thermal-transient and thermo-elastic-plas- tic stages may be uncoupled without penalty, as suggested by various authors (3-7). In (8), a coupled transient thermal and structural analysis of the temperature distribution and deformation was carried out. The transient temperature field calculated at the first stage provides the input for the thermo-elastic-plastic stage. The outputs of the latter stage, in the form of the transverse angular deformation and contraction forces, provide, in turn, the input for the elastic finite-