Crack growth modelling and probabilistic life assessment of coke drums operating under fatigue conditions

• Published in: International journal of pressure vessels and piping 2001-11, Vol.78 (11), p.1011-1020
• Main Authors: Church, J.M., Lim, L.B., Brear, J.M., Jarvis, P., Lant, R.P.D., Middleton, C.J.
• Format: Article
• Language: English
• Subjects: Applied sciences; Coke drums; Exact sciences and technology; Fatigue crack growth; Fatigue crack initiation; Fractures; Industrial metrology. Testing; Maintenance scheduling; Mechanical engineering. Machine design; Mechanical properties and methods of testing. Rheology. Fracture mechanics. Tribology; Metals. Metallurgy; Steel design; Steel tanks and pressure vessels; boiler manufacturing; Thermal stresses
• ISSN: 0308-0161; 1879-3541
• DOI: 10.1016/S0308-0161(01)00118-1

Coke drums are thin-walled pressure vessels that experience some of the most severe thermal cycling conditions of all the components on a refinery. The consequential high thermal stresses, applied in a cyclic manner, can lead to premature drum failure in the form of through wall low cycle fatigue cracking. Increasingly, coke drum operators are adopting risk based inspection and maintenance scheduling; utilising instrumentation for the on-line monitoring of strains and temperatures on the drums during cyclic operation as the technical basis for decision-making. The data collected can be analysed deterministically using classical solutions to enable the computation of mechanical and thermal stresses. Alternatively, the data can be treated statistically to form part of the input to a probabilistic assessment, which can be used to infer the cumulative probabilities of crack initiation and failure as functions of operational cycles, thus quantifying the likely risks associated with future operation. This latter approach enables both technical and economic factors to be reconciled when making decisions regarding future maintenance, inspection and replacement strategies. The approach has the advantage, amongst others, of reconciling potential variability of input data, for example, local material property variations and variable measured strain ranges, without being unnecessarily conservative.