Engineering critical assessments (ECA) are routinely used to provide defect acceptance criteria for pipeline welds. The fatigue crack growth evaluation as part of ECA involves the use of a safety factor applied to the number of cycles of a given stress spectrum. It has been routinely presumed that a safety factor value of half of the corresponding safety factor used for S-N fatigue should be used in the ECA. For example, this was suggested in DNV OS F101 earlier than the 2012 edition.
This paper presents a rationale and example calculations as to why a safety factor of 1.0 for fatigue crack growth is wholly appropriate for design stage calculations (for re-assessment when the various parameters are well defined then it could be appropriate to use a higher safety factor for the cycle counts). A probabilistic approach accounting for the statistical coincidence of misalignment, defect size, weld under maximum loading, etc is described in this paper. The probabilistic approach (based on fatigue crack growth) was compared with the S-N fatigue calculation and the deterministic ECA result. The results demonstrate that the probability of occurrence of fatigue failure using the probabilistic ECA approach with safety factor of 1 is far less than the DNV-OS-F101 requirement of 25 × 10−4 for medium safety class.
This conclusion presumes that in the standard ECA, a worst onerous selection is made from the range of input parameters, e.g. wall thickness, misalignment, defect location etc.