API Recommended Practice (RP) 1183 considers three levels of assessment. Its processes up through Level 2, which were considered viable for single peak dents with smooth profiles, are the focus of Part III of this four-part paper series. Parts I and II developed the background and considered the inherent assumptions and predictive viability of the RP’s practices regarding the deformation and strain capacity of dents during their formation. Trending and error analyses indicated scattered predictions along with unconservative errors that exceeded -100%. Part IIIB builds on the above foundation and on the concepts and data inputs detailed in Part IIIA. This paper (Part IIIB) extends this scope to consider the effects of pressure cycling on the integrity of dented pipes. About 100,000 predictive comparisons evaluated the current and proposed practices of the RP benchmarked against convergent Level 3 finite element analysis and also the full-scale test results. Error trending and analysis indicates that the lives predicted using the existing and proposed RP practices scatter significantly and can be very unconservative leading to life errors that exceed -500%. An alternative approach to life prediction was developed as a stopgap, which is shown to correctly predict both where cracking initiates and the resulting fatigue lives for all among full-scale tests for which the necessary details have been published. Part IIIB closes concerning RP 1183 with the observation that both the screening criteria and the fatigue assessment methods at Level 2 and below frequently lead to unconservative errors well in excess of 100% in rough balance with their conservative predictions: the worst unconservative prediction approached -2500%. Some guidance concerning the safe use of the practices of the 1 st edition of RP 1183 also is noted, along with many related conclusions. Part IV, currently in preparation more broadly assesses the viability of the numerical formulations and modeling aspects that underlie RP 1183.
Leis et al. (Fri,) studied this question.