Different load spectra and individual load peaks might substantially relax high residual stresses as well as induce compressive residual stresses in welded components and, consequently, affect the fatigue performance of these joints. Consideration of peak loads and resulting relaxation of residual stresses in fatigue analyses can substantially enhance the accuracy of life prediction. The aim of the current study is to experimentally investigate the fatigue strength of welded joints subjected to different levels of overloads and variable amplitude (VA) loads and to develop local fatigue assessment method to account for the relaxation of residual stresses via a mean stress correction using the 4R method. The 4R method applies a local stress ratio for the mean stress correction considering material strength, residual stresses, applied stress ratio of external loading and local weld geometry in elastic–plastic material behaviour. Fatigue tests were carried out on fillet-welded longitudinal gusset joints made of S700 high-strength steels under applied stress ratio R = 0–0.1. A mild strength steel (S355) and ultra-high-strength steel (S1100) were selected as reference steel grades for the fatigue testing to study the material strength effects. Numerical analyses were conducted to evaluate the fatigue notch factors using the effective notch stress concept with the reference radius of rref = 1.0 mm and theory of critical distance (TCD) using the point method. The experimental results indicated that a substantial improvement in the fatigue strength capacity can be claimed in the joints subjected to tensile overloads, particularly in the studied S700 and S1100 steels. The higher-level overload (0.8fy), corresponding to the nominal cross-sectional area, improved the mean fatigue strength of the welded joints manufactured of high-strength S700 steel by approximately 60%, while the lower overload (0.6fy) improved the mean fatigue strength by 20%. In addition, a use of equivalent nominal stresses for the joints subjected to VA loads resulted in conservative assessments when employing S–N curves obtained for the CA loading. The 4R method, via the local mean stress correction for individual cycles, provided higher accuracy for the fatigue assessments.
