A New Approach of Predicting Fractured Tight Sandstone Reservoir Permeability Based on Electrical Image Logging

Abstract Permeability is an important input parameter in Tight reservoir characterization and evaluation, precisely predicting formation permeability is indispensable. However, permeability prediction faces great challenge in tight sandstone reservoirs, empirical statistical methods, and nuclear magnetic resonance (NMR) based models lose their role due to complicated pore structure and the effect of methane gas (CH4) or hydrocarbon to NMR responses. In addition, fractures also play important role in improving tight reservoir permeability, whereas current logging responses cannot be used to characterize this improvement besides electrical image logging. In this study, to quantitatively characterize the improvement of fractures to filtration capacity in tight sandstone reservoir and accurately predict permeability, the Triassic Chang 63 Member of Jiyuan Region, Northwestern Ordos Basin is used as an example, a novel model of predicting permeability from electrical image logging is raised. In this model, the porosity frequency spectra are first extracted to characterize the pore structure of fractured tight sandstones. Afterwards, two parameters, which are defined as the logarithmic geometric mean value (φmv) and the golden section point variance (σg) of porosity frequency spectrum, are extracted to characterize the contribution of fractures to permeability. Comparing with the shape of porosity frequency spectrum, permeability φmv and σg, the quality of our target fractured tight sandstone reservoirs is quantified, and relationships among permeability, φmv and σg are established. High-quality reservoirs exhibit wide porosity frequency spectrum, high values of φmv and σg, and vice versa. Three parameters, which are formation total porosity, φmv and σg, are chosen to establish a novel fractured tight sandstone reservoir permeability prediction model. The involved input parameters in this model are calibrated by using the routine experiments of 35 core samples. Finally, we apply this model into field applications to consecutively calculate permeability in the intervals with which electrical image logging is acquired. Comparison of predicting permeability with core derived results illustrate that our raised model is usable in the Chang 63 Member of Jiyuan Region. The average relative errors between these two kinds of permeabilities is only 16.54% in 12 wells. This research gives a novel technique of calculating permeability in fractured tight reservoirs. It can avoid the effect of CH4 or hydrocarbon on conventional and NMR logging responses, and will play a great important role in unconventional reservoirs permeability prediction and formation characterization.