Evaluation of the effects of homogenizing matrix block sizes on the simulation of naturally fractured reservoirs

A vast amount of recoverable oil is buried in naturally fractured reservoirs. Due to the heterogeneous nature of matrix and fracture systems, there are many complexities in the simulation and production of these kinds of reservoirs. Fractures in naturally fractured reservoirs may be caused by various natural factors such as tectonic forces. Fracture distribution does not have a systematic pattern in naturally fractured reservoirs, and fractures are scattered across the formation. For the sake of simplicity, uneven distribution of fractures is not completely considered within practical reservoir simulation methodologies such as the commonly used dual-porosity approach. Using the dual-porosity methods, naturally fractured reservoirs are divided into simulation cells consisting of several matrix blocks. Against the nature of naturally fractured reservoirs, in the dual-porosity method, the fractures are often uniformly spread throughout the entire reservoir, and the sizes of matrix blocks inside the simulation cells are considered equal. For many years the assumption of homogenized matrix blocks in simulation cells has been used in the industry without proper evaluation of its effect on simulation results. This work examines the assumption of uniform matrix block size on the simulation results of naturally fractured reservoirs. An approach was proposed to create a simulation cell with randomly distributed matrix blocks to mimic the actual condition of naturally fractured reservoirs. Furthermore, realistic simulation approaches were utilized to investigate the oil recovery factor of different non-homogenized simulation cells during gravity drainage and imbibition processes. Also, the recovery factor performances of non-homogenized simulation cells were compared with simulation cells with uniform matrix block sizes created with different homogenization methods. An examination of the results revealed that a significant amount of inaccuracy in the simulation results could be caused by homogenization. Moreover, two types of homogenization approaches were used. In the first approach, a simulation cell with homogenized matrix block sizes based on the arithmetic mean of variable size matrix blocks was built. Homogenized matrix blocks in the second approach were built according to the arithmetic mean size of matrix blocks in different x, y, z directions. The first homogenization method resulted in less error compared to the second one. As compared with the gas invaded zone, simulation results of the water invaded zone are more error-prone. The findings of this work can help for better understanding the homogenization effect and prove that disregarding the heterogeneity of matrix blocks may lead to severe errors in the simulation results of naturally fractured reservoirs. The outcome also highlights the need for a more precise modeling and simulation approach for naturally fractured reservoirs.