Modelling Shallow Dents Using Local Regression Methods and Finite Element Analysis

Abstract

Modern in-line inspections can detect shallow dents in pipelines, with depths less than 2% of pipeline diameter. These dents are very common in thin-walled, small diameter refined and multiproduct lines, and frequently coincide with longitudinal welds and girth welds. Traditional dent assessment methods (such as the EPRG approach) can be conservative. Dents can have short predicted fatigue lives, but shallow dents are not known to be a major cause of pipeline failure, unless they are associated with a weld, a gouge, a crack, or severe pressure cycling. The conservatism affects both static failure assessments and fatigue assessments, resulting in high repair rates for shallow dents. This conservatism is partly due to: • Limitations of how the dent shape is modelled in the assessment methods; • Simplifications of the modelling of the stresses range; • Limitations of the calculation of strains in a dent based on inspection measurements; • Inability to model the changing cyclic stress range with changing dent shape. This paper shows that high resolution geometry inspection data contains irregularities which need to be filtered and smoothed. Advanced local regression methods are shown to give effective smoothing by removing errors but retaining the important elements of the real dent shape. The smoothed dent shape is used with the strain estimation methodology given by ASME B31.8 Appendix R, and an appropriate strain limit (based on likely weld quality), to assess whether cracking is likely to have initiated during dent formation. A methodology is then presented, based on Finite Element Analysis (FEA), which improves the accuracy of cyclic stress assessments of shallow smooth dents. The FEA model geometry is provided by the smoothed version of the measured dent shape. The pressure at which the dent shape was measured affects the calculated dent shape and stress as internal pressure varies: this effect is included in the model. The calculated cyclic stresses are used with S-N curves, such as those in BSI PD 5500, to estimate dent fatigue life. This methodology is then applied to 88 dents in two pipelines operated by ExxonMobil in the UK, using detailed high resolution geometrical in-line inspection data, comprehensive pressure cycle measurement data and enhanced dent assessment using the FEA method. It is concluded that this methodology can significantly improve the operator’s pipeline integrity strategy.