| Summary: | Reservoir Simulation is a very powerful tool used in the Oil and Gas industry to perform and provide various functions including but not limited to predicting reservoir performance, conduct sensitivity analysis to quantify uncertainty, production optimization and overall reservoir management. Compared to explored reservoirs in the past, current day reservoirs are more complex in extent and structure. As a result, reservoir simulators and algorithms used to represent dynamic systems of flow in porous media have invariably got just as complex. In order to provide the best solutions for analyzing reservoir performance, there is a need to continuously develop reservoir simulators and reservoir simulation algorithms that best represent the performance of the reservoir without compromising efficiency and accuracy.
There exists several commercial reservoir simulation packages in the market that have been proven to be extremely resourceful with functionality that covers a wide range of interests in reservoir simulation yet there is the constant need to provide better and more efficient methods and algorithms to study and manage our reservoirs. This thesis aims at bridging the gap in the framework for developing these algorithms. To this end, this project has both an educational and research component. Educational because it leads to a strong understanding of the topic of reservoir simulation for students which can be daunting especially for those who require a more direct experience to fully comprehend the subject matter. It is research focused because it will serve as the foundation for developing a framework for integrating custom built external simulators and algorithms with the workflow of the model builder of our reservoir simulation package of choice i.e. Petrel with the Ocean programming environment in a seamless manner for simulating large scale multi-physics problems of flow in highly heterogeneous flow of porous media.
Of particular interest are the areas of model order reduction and production optimization. In-house algorithms are being developed for these areas of interest and with the completion of this project. We hope to have developed a framework whereby we can take our algorithms specifically developed for areas of interest and add them to the workflow of the Petrel Model Builder.
Currently, we have taken one of our in-house simulators i.e. a two dimensional, oil-water five-spot water flood pattern as a starting point and have been able to integrate it successfully into the “Define Simulation Case” process of Petrel as an additional choice for simulation by an end user. In the future, we will expand this simulator with updates to improve its performance, efficiency and extend its capabilities to incorporate areas of research interest.
|
|---|
