Abstract
Offshore pipelines are installed by lay barges where the girth welds after weld completion and water cooling are inspected by AUT. It is common practice to develop the flaw acceptance criteria based on the DNV RP-F108 procedures which refer to BS7910 for the fracture mechanics and fatigue crack growth analyses required to estimate the critical flaw heights and lengths through an Engineering Critical Assessment (ECA). The critical flaw size is then corrected for the 95% confidence limit flaw under sizing error to develop the tolerable flaw size acceptance criteria used by the AUT inspectors to sentence the welds.
DNV RP-F108 contains an Appendix C providing guidance for ECAs for sour service pipelines requiring input of fracture toughness or resistance and fatigue crack growth rates determined experimentally by testing in simulated operation environmental conditions with respect to partial pressure H2S, CO2 and pH. Test conditions may also include inhibitors if continuous inhibition is applied in service.
The DNV RP-F108 describes a deterministic procedure to determine the tolerable flaw size for the AUT sentencing of the girth welds and the input to the ECA is worst-case parameters for the peak axial stress, fatigue loading during installation, on bottom empty, flooded and in operation stress associated with lateral buckling and vortex induced vibration associated with free spans. Likewise for the fracture toughness the lower bound initiation toughness value is applied for the analysis along with a conservative fatigue crack growth acceleration factor compared to data from air tests. This approach has been successfully applied for large number of sour service pipeline installation projects by DNV and various installation contractors since 2006 covering NACE MR0175/ISO15156 Regions 1, 2 and 3.
The DNV subsea pipeline standard DNV-ST-F101:2021 specifies target failure probabilities for different failure limit states. One of the limit states is loss of containment which may be caused by fracture of a girth weld, hence the annual target annual failure probability for loss of containment due to girth weld fracture at the end of life for the pipeline should not be exceeded. The current practice of performing deterministic ECAs does not provide a measure of the safety level associated with the flaw acceptance criteria applied for the pipeline installation, during service or at end of life. On the other hand, the target safety level for loss of containment specified in the standard has been calibrated based on structural reliability assessment (SRA). This paper presents an approach of SRA to perform probabilistic ECAs to assess the safety level of a given initial flaw size through the operation until end of life.
The methodology is applied to an actual gas pipeline project operating in Region 1 sour service environment based on the deterministically determined flaw acceptance criteria. The analyses showed that the probability of loss of containment by fracture and fatigue crack growth far exceeds the DNV annual target failure probability at end of life if applying all the worst-case assumptions used in a conventional deterministic ECA based on DNV RP-F108. However, it is shown that the target probability of loss of containment at end of life can be met if accounting for uncertainties in the loading, probability of having lateral buckles, location of flaws in the pipeline girth welds, peak stress locations, fracture toughness distribution, fatigue crack growth rate uncertainty, model uncertainty in the failure assessment diagram (FAD), tensile properties distribution, and the probability of AUT detection/non-rejection. Hence, the tolerable flaw sizes from the deterministic ECA performed as per DNV RP-F108 and Appendix C for sour service, can be shown by the SRA to meet the DNV annual target failure probability for loss of containment at end of life.
