Sub-Surface Cracking in Circumferential Seam Welds: The New Face of an Old Problem? Results of the Metallurgical Analysis of Damaged Welds

Abstract

Samples removed from several cracked piping and header circumferential seam welds have been examined destructively as part of an effort to understand why in the last few years there have been an increasing number of such seam weld failures at US power plants. To assist in this effort the information available regarding the design and operating history of a number of those welds has been evaluated where it was available; in addition, a limited review has been made of literature pertaining to cracking at seam welds in elevated temperature piping and headers. Certain features of the failures, such as the fact that the cracks in many cases have initiated sub-surface and that the vast majority of the cracks have been detected in shop-fabricated welds, have raised concerns that these failures represent a new and unique damage mechanism for which no satisfactory technical explanation currently exists. The information obtained to date does not support those concerns. Instead, the results of the destructive analyses of damaged girth welds are consistent with the general understanding of failures at welds in pressure parts operating at elevated temperatures: that is, failure in such welds can occur at times far shorter than the expected rupture time for unaffected base metal for the following reasons: • Welds are heterogeneous structures consisting of multiple discrete regions, the properties of which can differ significantly. In some of these regions the elevated temperature strength and/or ductility of the material can be substantially inferior to the “average” base metal, so that under certain loading conditions the rate at which damage will accumulate in weaker, less ductile regions can be orders of magnitude higher than in the stronger, more ductile regions. • The stress state at circumferential seam welds in elevated temperature piping and headers can be a complex admixture of mechanical and thermally-induced stresses imposed on the metallurgically varied structure. The extensive experience with Type IV or Type IIIa cracking in the CrMoV piping in Europe has demonstrated that under the influence of “normal” operating loads premature failure at weakened regions of the weld are to be expected. That experience also has shown that, in the absence of a very high bending load related to failure of the component support, sub-surface cracking is not unexpected. • The damage in the welds examined as part of this study was in all cases concentrated in portions of the weld that previous research has indicated can be either creep weak and/or creep brittle; other areas of the weld showed no signs of visible damage.