Implicit and semi-implicit techniques for the compositional petroleum reservoir simulation based on volume balance

Abstract

In reservoir simulation, the compositional model is one of the most used models for enhanced oil recovery. However, the physical model involves a large number of equations with a very complex interplay between equations. The model is basically composed of balance equations and equilibrium constraints. The way these equations are solved, the degree of implicitness, the selection of the primary equations, primary and secondary variables have a great impact on the computation time. In order to verify these effects, this work proposes the implementation and comparison of some implicit and semi-implicit methods. The following formulations are tested: an IMPEC (implicit pressure, explicit composition), an IMPSAT (implicit pressure and saturations), and two fully implicit formulations, in which one these formulations is being proposed in this work. However, the literature reports some intrinsic inconsistencies of the IMPSAT formulation mentioned. In order to verify it, an iterative IMPSAT is implemented to check the quality of the IMPSAT method previously mentioned. The finite volume method is used to discretize the formulations using Cartesian grids and unstructured grids in conjunction with the EbFVM (Element based finite volume method) for 2D and 3D reservoirs. The implementations have been performed in the UTCOMP simulator from the University of Texas at Austin. The results of several case studies are compared in terms of volumetric oil and gas rates and the total CPU time. It was verified that the FI approaches increase their performance, when compared to the other approaches, as the grid is refined. A good performance was observed for the IMPSAT approach when compared to the IMPEC formulation. However, as more complex stencils are used, the IMPSAT performance reduces.