Predicting weld distortion in the design of automotive components

Abstract

This project was set up in order to investigate whether it may be valid to use a simplified approach to weld distortion prediction in order to make an assessment of the distortions occurring in a welded structure. Distortion may manifest as problems associated with production, which increase costs and reduce profit. In order to validate the proposed prediction approach an experiment was developed to gain detailed information about distortions occurring in a series of simple welded joints. The joints and the parameters used for the experiment were based on the processes and applications of the case study company. The experiment was set up with the aim of evaluating the general magnitude and directions of distortion in bead on plate and butt weld joints to develop a database of distortion. The method for inspection of the experimental samples was to scan the parts using a 3D laser scanner to collect a detailed resolution point cloud that could be analysed. From the experimental results a number of key factors relating to the welded joints were found relating to material thickness, weld speed and penetration for the four different modes of distortion occurring. The experimental results were compared with the published data and equations presented by other authors, and some general agreements found, however, some differences were evident. In order for a designer to adjust the shape and form of a components design to counteract the distortions to achieve a nominal tolerance, based on these results, it was necessary to develop some new models specific to the materials and process variables of the case study company. Using an approach based on DoE software techniques, response surfaces for the experimental results were generated. This allowed equations to be developed for each distortion mode, which a designer could use to make predictions in the design phase to reduce risk from distortion.