An Embedded Grid-Free Approach for Near-Wellbore Streamline Simulation

Summary Reactive-transport phenomena, such as carbon dioxide sequestration and microbial enhanced oil recovery (EOR), have been of interest in streamline-based simulation (SLS). Tracing streamlines launched from a wellbore is important, especially for time-sensitive transport behaviors. However, discretized gridblocks are usually too large compared with the wellbore radius. Field-scale simulations with local-grid-refinement (LGR) models often consume huge computational time. An embedded grid-free approach to integrate near-wellbore transport behaviors into streamline simulations is developed, comprising two stages of development: tracing streamlines in a wellblock (a gridblock containing wells) and coupling streamlines with neighboring grids. The velocity field in a wellblock is produced using a gridless virtual-boundary-element method (VBEM), where streamlines are numerically traced with the fourth-order Runge-Kutta (RK4) method (Butcher 2008). The local streamline system is then connected with the global streamline system, which is produced with the Pollock (1988) algorithm. Finally, the reactive-transport equation will be solved along these streamlines. The algorithm presented for solving near-wellbore streamlines is verified by both a commercial finite-element simulator and a Pollock-algorithm-based 3D streamline simulator. A series of computational cases of reactive-transport simulation is studied to demonstrate the applicability, accuracy, and efficiency of the proposed method. Velocity field, time-of-flight (TOF), streamline pattern, and concentration distribution produced by different approaches are analyzed. Results show that the presented method can accurately perform near-wellbore streamline simulations in a time-effective manner. The algorithm can be directly applied to one grid containing multiple wells or to off-center wells. Furthermore, assuming streamlines are evenly launched from the gridblock boundary or ignoring transport in the wellblock is not always reasonable, and may lead to a significant error. This study provides a simple and grid-free solution, but is capable of capturing the flow field near the wellbore with significant accuracy and computational efficiency. The method is promising for reservoir SLS with time-sensitive transport, and other simulations requiring an accurate assessment of interactions between wells in one particular gridblock.