Workflows to Optimally Select Undersaturated Oil Viscosity Correlations for Reservoir Flow Simulations

Abstract

Undersaturated oil viscosity is one of the most important PVT parameters to be measured and/or predicted in a fluid sample. Since direct experimental measurements are expensive and time-costly, prediction methods are essential. In this work, viscosity data from more than five hundred fluid reports are utilized, and all correlation methods available in the literature and implemented in commercial software for reservoir and production engineering calculations, including fracked systems, are evaluated against the dataset. The results of this work are intended to set up workflows that give insight as to which method should be selected when running flow simulations, with emphasis on complex simulations such as in the case of EOR. The developed workflows provide the optimal choice of the viscosity correlation for the case of distinct viscosity ranges, as well as when overall performance is sought. A surprising result is that one of the oldest known correlations from the literature gives the best results (minimizes average absolute relative error) when tested against this large dataset. This may be attributed to the high locality that alternative correlations exhibit, which makes them generalize poorly.