Abstract
The severe global warming issues have driven the energy industry toward renewable resources and low-carbon emissions activities. In response, the petroleum sector has initiated several projects to store carbon dioxide (CO2) in underground reservoirs. This thesis utilizes available open-source software tools to model and optimize CO2 storage. Specifically, the project is based on two software programs: Flow, a reservoir simulator that includes options to simulate CO2 flow behavior in the reservoir, and FieldOpt, a field development optimization framework that optimizes engineering and geological parameters of reservoir simulation. This thesis aims to extend the FieldOpt software package to improve capabilities related to CO2 injection projects.
This thesis will consider the following topics related to modeling of CO2 storage: simulate fluid flow in the reservoir, analyze the post-simulation data, update the simulation case according to optimization algorithms, and re-simulate the model. Running FieldOpt would integrate all the processes above, using the simulator Flow for the reservoir simulation, and in an interactive optimization process finding a local optimum.
This project will extend the FieldOpt software to enable options necessary for the optimization of CO2 storage. The FieldOpt software is written in the objectoriented programming language C++. This thesis will therefore discuss variables, functions, classes, and modules deployed in the FieldOpt software package that are relevant for extensions towards CO2 storage. While extending the software by developing new features, current functionality should be kept intact, and the overall structure of the software source code will be maintained. As a result, The project broadens the software application to CO2 storage area. The main contribution is a new feature where the objective function can be calculated by an external code, e.g., a Python script. This enables an easy way for users to implement fit-for-purpose objective functions. This greatly increases the applicability and ease of dealing with customized reservoir optimization problems in FieldOpt.
To demonstrate the reliability and convenience of the new features, a synthetic CO2 model is built to work as a base case to be optimized. The FieldOpt would jointly optimize the well control and well placement for the CO2 storage with an injection well and production well. The newly developed optimization framework is tested using the genetic algorithm and the particle swarm optimization method. The result of these tests proves that the new implementation in FieldOpt enables a flexible way to study the CO2 sequestration optimization problem with customized objective function value calculation.
