Inherent possibilities and limitations of finite element modelling of defective girth welds

Welds unavoidably show defects, which can negatively affect the integrity of the entire structure and, worst case, result in a failure. Defects of a considerable size should therefore be detected, assessed and, if necessary, repaired. The assessment of a defect requires a procedure which allows a conservative estimation of the acceptability of the defect. To develop such procedure, both experimental and numerical research is performed. This paper describes the inherent possibilities and limitations of numerical research through finite element modelling, as compared to experimental research. Summarizing all arguments, it becomes clear that numerical research is a highly powerful tool, but a thorough experimental validation is of paramount importance. Moreover, some specific weld-related problems are highly difficult to address, namely the presence of stable crack growth and material heterogeneity. More research is needed to achieve a description of these phenomena, under a set of conservative assumptions