A novel method to couple well bore flow to reservoir flow to understand the impact of a well bore failure during CO2 sequestration operation

Abstract

One of the major concerns of geologic sequestration is consequences of a failed well bore and subsequent release of CO2. The release can be slow or catastrophic depending on the mode of well bore failure. It is important to predict the release characteristics of a failed well bore in order to develop monitoring plans, to understand potential impact and to develop strategies to mitigate the leak. In order to understand the total amount of CO2 that can escape due to release through failed well bores, it is necessary to perform numerical simulations coupling well bore flow to reservoir flow. These are complex simulations, requiring solution of coupled equations describing mass and heat transfer processes occurring between the well bore and surrounding rock. In addition, it is important to capture the CO2 phase change behavior as it escapes through well bores as it can impact the flow behavior through the well bore. Traditionally, adding well bores to reservoirs to capture short and long-term dynamic behavior in the vicinity of the well bore has required refining the computational grid. In addition, to accurately capture the thermodynamic effects within the well bore (such as density changes) requires representing the well bore in multiple segments. The traditional approach results in an extremely large computational grid for a reservoir with multiple well bores. In addition, the traditional approach has limited flexibility when introducing new well bores to an existing computational grid. We have developed a computationally efficient method to couple well bores to an existing numerical grid using FEHM (Finite Element Heat and Mass), the Los Alamos National Laboratory’s numerical code for simulating heat and mass flow in porous media. FEHM uses a control volume finite element approach to solve numerical equations for heat and mass flow. The equations are generated using an unstructured grid approximation and are solved using a multiple degrees of freedom Krylov space method. This approach has the advantage where new connections to an existing (primary) grid can be easily added or modified, without needing to regenerate a completely new grid or create a large computational grid. We use this capability to add well bores to an existing grid. The well bore is added by connecting well bore nodes to the primary computational grid nodes. The addition is done in a way that does not require modifying existing grid block connectivities. The equations representing the physics of well bore flow are coupled to the reservoir flow equations through an implicit formulation. The approach can be used to connect multiple well bores, each with multiple segments, at any desired locations within an existing grid. This approach does not lead to large computational grids and results in a computationally efficient method to simulate coupled well bore and reservoir flow. In this paper we provide details of the above-mentioned approach. We also provide results of the simulations comparing the solutions obtained using the new approach to the solutions obtained using the analytical approach as well as the numerical approach where a highly refined grid is used to represent the well bores.