The need to predict reservoir fluid compositional changes under varying pressures and depletion processes arises because the hydrocarbon recovery of some types of reservoirs is strongly composition-dependent. Since hydrocarbon component material balance equations must be solved in compositional reservoir simulation, both computer memory requirements and simulation run time are greatly increased.

This paper presents a series of technologies, including a Fully Automatic Regression Technique, a Best Lumping Scheme, and a Fully Automatic Reservoir Fluid Characterization procedure, to reduce the number of components needed to characterize a reservoir fluid using an Equation of State (EOS). As a result, the compositional reservoir simulation run time can be greatly reduced without reducing the required accuracy of the match with available laboratory data. Applications indicate that a significant speedup has been achieved using fewer components to model the reservoir production processes while the predicted reservoir performance is equivalent to that achieved using many more components.

You can access this article if you purchase or spend a download.