Abstract
Small bore austenitic stainless steel pipework is used in a number of nuclear plant systems. Many of these locations are subjected to large thermal shocks and therefore have high fatigue usage factors. Their justification therefore often includes a fatigue crack growth and fracture assessment, for which a key input is the residual stress associated with the welding process, in UK assessments these are typically taken from the R6 compendium.
A common process used for these welds is manual tungsten inert gas welding, due to access difficulties each pass is usually completed in two halves. The stop-start locations for each weld run are sometimes stacked, especially in horizontal pipe runs where each weld operation starts at the bottom of the pipe and progresses upwards. The stack up of stop-start locations is likely to lead to considerable circumferential variation in weld residual stress, potentially resulting in stresses that locally exceed the R6 profiles.
This paper presents results from a series of FE models for a single small bore pipe weld. The simulated weld is a 3-pass manual TIG weld with an EB insert in a 2 inch (50 mm) nominal diameter pipe. Both 2D and 3D models were run. The results of the modelling are then compared with measurements of weld mock-ups of the same weld (both with and without the stop-start stack-up).
The results show that, local to the assumed stop location the predicted stresses do exceed even the R6 level 1 profile (a membrane stress equal to the 1% proof stress of the material). However, the locally enhanced stresses drop off quickly away from the peak location, so for defects of a size that may be a concern for a defect tolerance assessment, the R6 Level 1 and 2 profiles remains appropriate or bounding.