Fatigue of high quality pipe girth weld

Abstract

This study aims to investigate the influence of weld toe grinding and center eccentricity to high quality pipe girth welds. A series of 3D FE models with different weld toe geometries and center eccentricity subjected to bending moment were modeled with finite element software Abaqus Standard 6.8-2. Analytical expressions for stress concentration factors were established. An overview of up-to-date possibilities for post weld improvement techniques is also included.

In general, the fatigue performance of welded details can be improved by good design practice and high quality fabrication. However, as an alternative to strengthening the structure, post weld improvement techniques can be applied to increase the fatigue strength of low category details of new structures or to repair or upgrade existing structures. The improvements in fatigue strength obtained by peening treatments are among the highest reported. Compounding of improvement techniques, i.e. a combination of a weld geometry improvement method and a residual stress method has proved to give good results.

A great deal of effort was used to establish the most applicable 3D FE models for the series of analyses. The model approaches consists of several separate parts which are tied together by surface-based tie constraints. Dividing the FE model into smaller parts made it possible to use structured meshing techniques and keep the number of DOF within feasible limits for computation. A good mesh of hexahedral elements is known to provide quality solution at less computational cost. At first a very fine mesh at the notches was tried, however this was difficult to achieve with hexahedral elements as this type of element shape proved to be unsuitable. Therefore tetrahedral elements were used for the swept and lofted parts. 10-node modified quadratic tetrahedron elements (C3D10M) were successful in describing the notch stresses. Feasible limits for computation were exceeded long before a distinct convergence of the notch stress was obtained. It is expected that the notch stress (S33) will continue to slowly decrease as the mesh gets denser. Using edge seeds smaller than 0.4 mm3 extended the computational time but did not improve the FE analysis results significantly. It was concluded that the maximum notch stress based on an average elements size of 0.4 mm3 was suitable for the series of analyses.

FE stress analyses of pipe girth welds improved by toe grinding were performed. The girth weld and the notches were modeled using swept features. The maximum stress (S33) was located at the bottom of the upper notch and decreased with greater vertical distance from the top. The vertical distance of the area which contains maximum S33 corresponds with an angle of 33.1° from the top. The S33 at distance of 1D (OD pipe) between the middle of the weld and the constraint is not affected by the stress noise at the notch.

FE stress analyses of misaligned pipe girth welds improved by toe grinding were performed. The part which represented the pipe girth weld and the notches was modeled by adding solid loft features to a swept part. The model approach was successful in modeling the complex weld geometry. Lofted features provided surprisingly good mesh control and responded very well to denser local mesh by not reducing refinement within the first six element layers from the notch surface. The maximum stress (S33) was located at the upper right notch, and decreased with greater vertical distance from the top. The vertical distance of the area which contains maximum S33 corresponds with an angle of 28.1° from the top. The S33 at distance of 2D (OD pipe) between the middle of the weld and the constraint is not affected by the stress noise at the notch.

The results of the analyses show that some of the variables have a stronger influence of the notch stress than others. Increasing the height of the weld 10% does not influence the SCF as much as a 10% increase of the radius of the notch or a 10% increase of the vertical range of the notch. Increasing the radius provides a lower SCF, while increasing the vertical range and the hight of the weld increases SCF. A combination of large eccentricity, small radius and large vertical range of the notch is the most unfavourable situation with regard to fatigue strength. Second degree polynomials were used in the process of constructing curves and mathematical functions of SCF and seem to be a good approximation.