Residual stress within stainless-clad bimetallic steel welded box sections

Abstract

Stainless-clad (SC) bimetallic steel is a new high-performance steel which is composed of a substrate layer (CM steel) and a clad layer (stainless steel), and they are bonded metallurgically together. It is beneficial to both structural performance and corrosion resistance together with a high cost efficiency. This paper presents an experimental programme to investigate and model the residual stresses within SC bimetallic steel welded box sections. Eight box sections with various widths, thicknesses and clad ratios were fabricated and tested with sectioning method. A total of 494 strip specimens were obtained to quantify both magnitudes and distributions of welding-induced residual stresses within these sections. In each section, the maximum compressive stress is significantly smaller than those reported previously, while the compressive stress within the clad layer is much higher than the one within the substrate layer. A simplified model which ignores variations of magnitudes along the thicknesses was proposed. The distribution regions of this multi-stepped model are determined and the compressive residual stresses were found to be dependent on both widths and thicknesses of the sectional component plates. Moreover, deviation of the compressive residual stresses within the clad and the substrate layers was observed in this study, and it was demonstrated that this deviation depended significantly on the clad ratio of the bimetallic steel plates. A layered model was further established after considering these differences between these residual stresses within the two layers, in which the deviation of the compressive residual stresses within the clad and the substrate layers is found related to both the width-to-thickness ratio and clad ratio of the component plate. Two models proposed herein are shown to agree well with the measured residual stress distributions. Furthermore, numerical models were established to assess accuracy of these two models in simulating column buckling, and it showed that both the simplified model and the layered model were applicable to simulate column buckling behaviour.