When samples extracted from butt welds in 50mm plate gave fatigue results which were not as expected, residual stress measurements were used to provide an explanation.
A TWI member company was carrying out an investigation into fatigue crack growth rates in weld metals. A selection of test weldments had been prepared - all butt welds in plate around 50mm thick, made with different welding processes and procedures.
To investigate fatigue crack growth rates at different depths within the weld metal, small transverse test specimens were cut from the outer surfaces and centres of each weldment. These specimens were similar in size and shape to the standard 10mm x 10mm Charpy impact test specimen.
Fatigue crack growth rates found in the tests were higher in the specimens cut from the centre of the weldment than in the near-surface specimens.
It was known that residual stress in the weldments could influence the fatigue crack growth rate, but the precise through-thickness distribution in each weldment was unknown. Furthermore, it was known that removal of small test specimens from a weldment often causes a redistribution of residual stress in the specimens, because of the release of stored energy and restraint of the surrounding material.
One of TWI's residual stress measurement techniques was ideally suited to this investigation. The 'block removal, splitting and layering' method for determination of through-thickness residual stresses in weldments is a destructive method of measurement, which works as follows:
1.A block is marked out on the welded panel, and strain gauges are mounted on both outer surfaces.
2.The block is sawn out of the weldment, and split on the plane of mid-thickness.
3.Each half-block has thin layers machined from it, starting at the mid-plane and working towards the gauged outer surface.
4.Strain changes are recorded after each cutting operation. These data allow the residual stress in each layer, and hence in the complete weldment, to be calculated using simple beam theory.
In this investigation, the majority of weldments tested revealed a typical transverse through-thickness residual stress distribution of tensile yield magnitude near the surface and compressive yield magnitude in the centre of the weldment (Fig.1).
The data were also used to calculate the redistribution stresses in the small specimens extracted from the weldment. Those extracted from the surface of the weldment, where the original residual stresses had been highly tensile, were found to have compressive residual stress at their surfaces after removal (Fig.2). Conversely, the specimens extracted from the centre of the weldment, where the original residual stresses had been highly compressive, were found to have predominantly tensile stresses at their surfaces after removal (Fig.3).
The derived residual stress distributions in the Charpy-size specimens explained the observation that increased fatigue crack growth rates were found in the specimens extracted from the centre of the weldment, despite the fact that this region would contain compressive residual stresses in an intact weld.
For information about TWI's capabilities please contact us.