HIGH RESOLUTION PREDICTION OF GAS INJECTION PROCESS PERFORMANCE FOR HETEROGENEOUS RESERVOIRS

摘要

This final technical report describes and summarizes results of a research effort to investigate physical mechanisms that control the performance of gas injection processes in heterogeneous reservoirs and to represent those physical effects in an efficient way in simulations of gas injection processes. The research effort included four main lines of research: (1) Efficient compositional streamline methods for 3D flow; (2) Analytical methods for one-dimensional displacements; (3) Physics of multiphase flow; and (4) Limitations of streamline methods. In the first area, results are reported that show how the streamline simulation approach can be applied to simulation of gas injection processes that include significant effects of transfer of components between phases. In the second area, the one-dimensional theory of multicomponent gas injection processes is extended to include the effects of volume change as components change phase. In addition an automatic algorithm for solving such problems is described. In the third area, results on an extensive experimental investigation of three-phase flow are reported. The experimental results demonstrate the impact on displacement performance of the low interfacial tensions between the gas and oil phases that can arise in multicontact miscible or near-miscible displacement processes. In the fourth area, the limitations of the streamline approach were explored. Results of an experimental investigation of the scaling of the interplay of viscous, capillary, and gravity forces are described. In addition results of a computational investigation of the limitations of the streamline approach are reported. The results presented in this report establish that it is possible to use the compositional streamline approach in many reservoir settings to predict performance of gas injection processes. When that approach can be used, it requires substantially less (often orders of magnitude) computation time than conventional finite difference compositional simulation.