Dimensional Analysis and Validity of Uniaxial Residual Stress Distribution for Welded Box Sections

This paper investigates the residual stresses induced by metal inert/active gas (MIG/MAG) welding in normal strength steel box sections, focusing on the validity of uniaxial residual stress assumption. Advanced manufacturing simulations are conducted using deterministic uncoupled transient thermomechanical analysis with a double-ellipsoidal heat source model, employing 8-node solid elements and material models calibrated for extreme temperatures per EN 1993-1-2. A comprehensive parametric analysis investigates the effects of primary welding variables, such as heat source power and welding speed, alongside geometric parameters of the heat source model using random Latin hypercube sampling technique in the analyzed parameter set. The relationship between the size and shape of the characteristic isotherms, i.e., the aspect ratio and the Rosenthal number, underscores that the analyzed welding heat sources are in the fast regime with the validity of uniaxial residual stresses based on the analytical assumption (minimal values are AR = 9.94 and Ro = 30.47). The validity and limitations of uniaxial residual stress assumptions for 59 welded and 51 heated box sections are critically evaluated by using the finite element model-based stress triaxiality parameter. Results confirm that longitudinal residual stresses dominate typical MIG/MAG welding applications, supporting the application of uniaxial residual stress models in advanced structural design by neglecting in-plane and through-thickness residual stresses. Conversely, three-dimensional residual stress state dominates under conditions such as preheating or thermal straightening.