Effect of Welding Residual Stress and Distortion on Ship Hull Structural Performance

Abstract

The finite element method is used to investigate the effects of welding-induced residual stress and distortion on the strength and behaviour of ship hull structures. A finite element welding simulation consisting of sequentially coupled transient thermal and nonlinear structural analyses is used to predict the three-dimensional residual stress and distortion fields in welded stiffened plates. Three types of stiffener commonly used in commercial and naval applications are considered. The welding simulation is followed by a 'shakedown' analysis to study the possibility of residual stress relief caused by cyclic loads. The strength and behaviour of stiffened plates under axial load is characterized by normalized plots of average axial stress versus axial strain, commonly referred to as load-shortening curves. These curves are used to evaluate the effects of welding-induced residual stress and distortion on stiffened plate behaviour with and without considering stress relief by shakedown. Load-shortening curves generated by finite element analysis are also compared with load-shortening curves produced using analytical methods including those prescribed in ship structural design standards published by the International Association of Classification Societies (IACS). To conclude, a hull girder ultimate strength analysis is carried out using Smith's method with load-shortening curves generated by several different methods. Results indicate that welding-induced residual stress and distortion decrease the ultimate strength of flat-bar, angle, and tee-stiffened plates investigated in this study by as much as 17%, 15% and 13%, respectively. Stiffened plate ultimate strength values calculated using IACS common structural rules agreed reasonably well with results from numerical models in most cases. There was however, a significant discrepancy between the numerical load-shortening curves and the IACS curves in the post-ultimate regime, where the IACS curves overestimated the post-ultimate strength of stiffened plates by as much as 30%. To investigate stress relief by shakedown, axial stresses of 25% and 50% of the yield stress were applied and residual stresses were reduced by approximately 20% and 40%, respectively. In some cases, these reductions in residual stress led to increases in stiffened plate ultimate strength as high as 7%. Analysis of a box girder using load-shortening curves from a finite element model including residual stresses and distortions predicted by welding simulation predicted a bending moment capacity within 2.7% of the experimentally measured value. Using load-shortening curves from the IACS common structural rules, the ultimate strength was overestimated by 17%.