Implementation and Benefits from Novel Algorithm for Complex Fractured Carbonate Reservoirs in Abu Dhabi Onshore Field

Abstract Objectives/Scope This work outlines the implementation and evaluation of a novel reservoir simulation algorithm within a complex, highly heterogeneous, and fractured carbonate reservoir situated in the Middle East. The reservoir, with over 40 years of production history and hundreds of wells, exhibits confirmed production from both matrix and fracture zones, with fracture lineaments modelled based on seismic and image logs. Currently the reservoir is modelled as dual porosity and dual permeability with the reservoir simulation employs a fine grid comprising 42 million matrix cells. Methods, Procedures, Process Originally, the fine-grid model was evaluated using the commonly used conventional reservoir simulation solver which is Fully Implicit (FI) that results in high running time and was not capable of handling the complexity, size, and challenges of the reservoir. A state-of-art solver technology known as Multiscale solver was tested on this high-resolution dual porosity dual permeability model to understand the difference in running time and performance compared to the conventional solver. Results, Observations, Conclusions The Multiscale solver shows significant performance enhancements, achieving an impressive fourfold acceleration in simulation time compared to conventional solver without any hardware update. This acceleration not only facilitates the exploration of fine-scale models to enhance reservoir management quality and produce more reliable forecast plans but also expedites the identification of uncertain parameters for improved reservoir understanding and characterization, thereby accelerating the history matching process. Novel/Additional Information The Multiscale solver is based on an alternative formulation of the reservoir equations, known as the Sequential Fully Implicit (SFI) formulation. It divides the displacement process into two parts: one system of equations for the reservoir pressure, and another system of equations for the transport of fluid components (in other words, the phase saturation / molar fractions). The pressure equations are near elliptic. It can be difficult to solve in parallel because of the requirement for global communication and sophisticated linear solvers. In contrast, the transport equations are near hyperbolic and strongly non-convex. But the linear systems are usually easier to solve in parallel, and simpler linear solvers can be applied efficiently. By tailoring the solution process to each equation and the interactions between them, the SFI formulation can perform better than the conventional Fully Implicit method. The improvements in the simulation solvers played an important role in running high-resolution models that capture geological concepts and high degree of heterogeneity that were not able to capture before. This paper sheds light on this novel solver from both theoretical and application points which is considered a major step change in reservoir modeling.