Abstract
Probabilistic methods can, by quantifying the effects of uncertainties, provide a much more realistic evaluation of remaining lifetime than can be achieved with conventional deterministic models. Using data from an actual process plant, lifetimes for catalytic reformer tubes were calculated. The tubes were made of an HP-45 Niobium modified alloy that had been in-service for approximately 10 years. A strain based model linked to a non-linear finite element program was utilized to account for the thermal stresses when evaluating the tube damage due to creep. Various start/stop procedures were evaluated and the results quantified in terms of tube damage. A sensitivity analysis was also performed upon the input parameters which have been ranked according to importance. The calculated results were then compared with the actual operating and inspection experience of the unit. Since probabilistic methods alone cannot make a fundamentally inferior model correct, a form of verification was developed by comparing the calculated amount of tube damage with past inspection results. In particular, the model was able to accurately predict the number of C-rated tubes from inspection.
