This thesis work is concerned with the mechanical response of welded steel structures, which are distortions, residual stresses, and fatigue. The accuracy of fatigue assessment, distortion and residual stress analysis using Computational weld mechanics (CWM) is focused. The following studies are performed; welding simulations of residual stresses and distortions, weld quality estimation, evaluation of local stress-based fatigue strength assessment methods.

The following CWM methods: thermo-elastic-plastic, inherent strain (local-global), and sub structuring have been implemented both on T-type fillet weld and butt weld specimens to estimate angular distortion and residual stresses. In regard to large welded structures, the CWM methods using lumping method, together with prescribed temperature method, is implemented on a welded box structures to estimate welding residual stress state. The welding distortion analysis has been carried out on a large full scale beam structure experimentally and numerically using CWM techniques such as the inherent strain (local–global) method and the shrinkage method, together with the lumping approach. The accuracy of CWM methods is studied, and improvements are proposed.

A probabilistic model is proposed to estimate the probability of a targeted weld penetration depth. The uncertain process parameters are voltage, current, travel speed, and torch travel angle which were studied based on an experimental investigation. The weld penetration depth is evaluated from macrographs using a digital tool developed in MatLab. The epistemic measurement uncertainty related to this evaluation is quantified and incorporated in the probabilistic model.

Monte Carlo simulation is implemented to consider the weld geometry variations in the ENS methods. The stochastic variability in toe radius, toe angle and leg length are considered. The influence of weld geometry variations on the ENS methods using deterministic and stochastic SCF distribution is assessed.

The Hot Spot Stress (HSS), 1-mm stress (OM), Theory of Critical Distances (TCD), Stress Averaging (SA), and Effective Notch Stress (ENS) methods are evaluated for cover plates and T-joints subjected to axial and bending loading, in terms of accuracy and reliability. The evaluation is based on fatigue test data extracted from the literature and carried out in this study. The fatigue design curves applicable for T-joints under bending are discussed, which can be used in the TCD method and SA method.

The studies above contribute to increasing the accuracy in the estimation of residual stresses and distortions using simplified CWM methods, achieving a required reliability level in manufacturing, and improving accuracy and reliability of local stress-based fatigue assessment methods.