OSCAR GRANBERG

Sand erosion is the cause of widespread structural damage in offshore pipe systems and it is often associated with large costs. Correlative and empirical methods are often used to predict erosion rates. Such procedures, however, tend to yield overly conservative estimates. In this thesis, a Lagrangian CFD model has been compared with some of the main correlative methods for predicting erosion in gas streams for pipe bends and blinded tees. It is found that the correlative methods differ by several orders of magnitude. The DNV RP O501 correlation under-predicts erosion by more than one order of magnitude for low density fluid mixtures at velocities below 10 m/s. Maximum erosion rates are obtained at intermediate sand particle diameters both for the pipe bend and blinded tee. Furthermore, an Eulerian model for high particle loadings, derived for use in fluidized-bed combustors, has been adapted to predict erosion on steel grades. The model reproduces similar erosion patterns, as well as velocity and particle diameter dependencies, as correlative methods suggest. The magnitude of erosion has not been reproduced, presumably due to an erroneous proportionality constant. Eulerian methods are associated with inherent losses of vital information, such as the particle impingement angle and impact phenomena. We believe that the outlook is generally poor for accurate wear predictions in the Eulerian framework and capturing of effects such as lowered erosion efficiency in dense mixtures