Numerical Methods to Predict Distortion in Welded Components
Numerical methods have been established to simulate welding processes. Of particular interest is the ability to predict residual stress fields. These fields are often used in support of structural integrity assessments where they have the potential, when accurately characterised, to offer significantly less conservative predictions of residual profiles compared to those found in assessment codes such as API 579, BS7910 and R6. However, accurate predictions of residual stress profiles that compare favourably with measurements do not necessarily suggest an accurate prediction of component distortions. This paper presents a series of results that compare predicted distortions for a variety of specimen mock-ups with measurements. A range of specimen thicknesses will be studied including, a 4mm thick DH-36 ferritic plate containing a single bead, a 4mm thick DH-36 ferritic plate containing fillet welds, a 25mm thick 316L austenitic plate containing a groove weld and a 35mm thick esshete 1250 austenitic disc containing a concentric ring weld. For each component, distortion measurements have been compared with the predicted distortions with a number of key features being investigated. These include the influence of ‘small’ vs ‘large’ strain deformation theory, the ability to predict distortions using simplified analysis methods such as simultaneous bead deposition and the influence of specimen thickness on the requirement for particular analysis features. The work provides an extremely useful insight into how existing numerical methods used to predict residual stress fields can be utilised to predict the distortions that occur as a result of the welding fabrication process.