| With the explosion of computational power and development of new and advanced simulation techniques, reservoir simulation has evolved into a mature technology. Today's simulator is more robust, more complex and more useful than what was available twenty years ago. Under these circumstances, a good design for the simulator is essential. So the first objective of this research is to develop a new General Purpose Research Simulator (GPRS), which can be used later by a large research group to preserve and enhance the work of several investigators.; Compositional simulation is the major focus of this research. It is more complex than black-oil simulation due to the large number of components and various issues related to flash calculations. Because of this complexity, a large number of compositional models have been proposed. They use different equations, variables and implicit levels. The second objective of this research is to evaluate the performance of different compositional models. To do this in a consistent manner, a new General Formulation Approach is developed to derive any kind of model.; The IMPSAT (implicit pressure and saturations and explicit mole fractions) model has been discussed in the literature. Here we investigated this method in detail, including its characteristics, stability, construction and performance. Besides this, a series of new IMPSAT based AIM (adaptive implicit) models are also proposed, and their performance is compared with the performance of the traditional AIM model.; Various models and techniques of GPRS have been tested and validated, including the black-oil model, compositional model, unstructured grid handling and multi-point flux approximation. The performance of different compositional models has been evaluated using a wide range of problems. We found that the natural variables (pressure, saturations and component mole fractions) are good for the fully implicit (FIM) model, while the overall variables (pressure, overall component mole fractions) are good for the IMPES model. We have also demonstrated that the IMPSAT model and the IMPSAT based AIM models have very good performance for a wide range of problems, and they are more efficient than the traditional models. |
