Choosing an appropriate technique for upscaling the permeability of the discreet fracture network (DFN) model is vital for maximizing recovery from naturally fractured reservoirs (NFR). Flow -based upscaling is accurate, yet it is computationally expensive. Analytical (i.e. Oda method) upscaling is computationally efficient; however, it is only accurate for well-connected fractures. The objective of this paper is to analyze the performance of the newly developed Oda corrected method which addresses issues associated with previous upscaling methods.
This research was executed by using a commercial numerical simulator and by using data set from the Teapot Dome Reservoir. Furthermore, DFN modeling was used to generate different realizations for the fracture network. Consequently, sensitivity analysis was performed through a realisti c uncertainty quantification to generate a base case and 6 different DFN realizations. The main parameters used for this study are fracture length and intensity. Afterwards, the fracture permeability corresponding to each DFN realization was upscaled using the above-mentioned methods. Finally, differences between the upscaling methods were evaluated and analyzed using flow-based upscaling as the criterion.
Indeed, the analysis revealed that the new Oda corrected method can calculate the equivalent permeability tensor with adequate accuracy. However, it overestimates the permeability when fracture networks are below the percolation threshold and/or when fractures length is large. Hence, this method is recommended for networks with moderate to high intensity. Furthermore, it has been deduced that fracture length has a great impact on the connectivity of fractures, albeit its effect on permeability is limited by fracturing density. Additionally, it has been found that the length of fractures has an immense impact on the anisotropy ratio and control the occurrence of water bypassing, which were not captured by the Oda method.