A Simulator for Predicting Thermal Recovery Behavior Based on Streamline Method

Abstract

Abstract The general tendency in simulation technique of thermal recovery processes has been to use the fully implicit finite difference method to solve the mass and heat transport. The method has been proven favorable to handle rapidly changing properties and complex physical mechanisms encountered in the processes. However, this method is limited in the grid size due to restrictions in computational power. In addition, the method could also erroneously predict recovery performance if care is not taken in the choice of numerical scheme. For example, when the mobility ratio of the displacement is unfavorable, calculations often result in exaggerated global sweep and may adversely affect reservoir performance prediction. Local grid refinement is a natural process to handle this situation but requires a great amount of memory and high computational cost. This becomes inefficient when multiple simulations are required to assess a wide range of reservoir development scenarios. An alternative approach to solve the problem is to use the streamline method. Because fluid transport occurs along streamlines rather than between discrete grid blocks on which the reservoir pressure field is solved, the problems associated with the fully implicit finite difference method are minimized and the simulations can run much faster. This is supported by the results presented in this paper. We applied the developed thermal streamline simulator to a hot water-flooding process. Test simulations were performed in the two-dimensional (2D) areal quarter five spot pattern. The heavy oil reservoir is assumed to have undergone cold water injection. Water at the initial conditions is about 45 times more mobile than oil, which leads to unfavorable displacement. The results of simulations are compared with those from the fully implicit method using the 5-point and 9-point finite difference schemes. Although simplified, the simulation demonstrates the main features of our simulator.