Evaluation and Comparison of Reservoir Simulator Performance in Waterflooding Scenarios Using MATLAB Reservoir Simulation Toolbox (MRST) and Computer Modelling Group (CMG)

Abstract

Reservoir simulation plays an important role in optimizing field development strategies, particularly in waterflooding scenarios. This study provides a comprehensive evaluation of reservoir simulation performance by comparing the open-source simulator MATLAB Reservoir Simulation Toolbox (MRST) and the commercial simulator Computer Modelling Group (CMG) on various sensitivity variables and computational time efficiency. Sensitivity analysis was performed by varying the grid size, applying linear, five-spot, and inverted five-spot injection patterns, and conducting tests at several injection pressures. The reservoir fluid model was built on both simulators, resulting in a 1.5% difference in Original Oil in Place (OOIP) values, indicating that the model setup was nearly identical on both simulators. The results of the study indicate that increasing grid resolution from 300 to 14,700 blocks does not always smooth the slope decline of reservoir pressure and oil production rates. The five-spot and inverted five spot patterns improved the recovery factor compared to the linear pattern, with the inverted five-spot pattern showing more effective fluid sweeping. At low injection pressures, numerical instability was observed in MRST. The instability consisted of fluctuations in the injection rate, unstable water breakthrough, and oscillations in the oil production profile and Gas Oil Ratio (GOR) values. Part of this instability may have been caused by the solver characteristics, unadaptive time-step adjustment, and challenges in building the fluid model. In terms of computational efficiency, CMG performs significantly more efficiently, completing simulations over 77 times faster than MRST. Overall, this study demonstrates that MRST is more suitable for academic research and new method development, whereas CMG remains the preferred option for large-scale reservoir simulation applications with high time efficiency and numerical stability requirements.