Predicting Permeability in Tight Sandstone Reservoirs: An Integrated Approach Using Nano-Scale CT Scanning and Nuclear Magnetic Resonance (NMR) Logging

Abstract

Accurately predicting permeability in tight sandstone reservoirs remains a formidable challenge due to their complex pore systems, characterized by nano-to micro-scale pores and strong heterogeneity. Conventional methods and empirical models often fail to capture the intricate pore-throat connectivity, leading to significant uncertainties in reservoir evaluation and development plan formulation. This study aims to quantitatively characterize the full-range pore structure, and develop a new permeability prediction model by integrating nano-scale X-ray computed tomography (nano-CT) with nuclear magnetic resonance (NMR) logging. The research workflow is executed in four stages. First, 87 representative core samples were drilled from the Wenchang Formation in LF Depression of the eastern basin of the South China Sea and applied for nano-CT scanning, and the porosity values for three different pore sizes, such as nanoscale porosity, microscale porosity and submillimeter-scale porosity, and their respective proportion in total porosity were obtained. Concurrently, laboratory NMR measurements were conducted on the same samples to acquire the NMR T2 spectra and permeability values. Second, relationships among pore space at different scales and rock permeability are analyzed, and we find that the key factor controlling permeability in tight sandstones is not total porosity, but the microscale porosity (φmic). The contribution of nanoscale and submillimeter-scale porosities to permeability can be ignored. Meanwhile, we also find that the NMR T2 geometric mean (T2gm) is also heavily associated with permeability. Third, based on the heavy relationships among φmic, T2gm and permeability, we establish a novel permeability prediction model, and the form of this model is similar to the classic Schlumberger Doll Research center (SDR) model. Finally, to consecutively acquire φmic, we combine nano-CT and NMR data to extract two T2cutoffs, and they classify the T2 distribution into three parts. Afterwards, the φmic and T2gm is calculated from the NMR T2 spectrum, and our raised model can be used to predict permeability. We extend the raised model into field application in the LF Depression, the NMR logging is processed to obtain the values of φmic and T2gm, and consecutively permeability curves are acquired in several wells. Comparisons of predicted permeabilities with core-derived results illustrate the reliability of our raised model. Once this model is applied in a tight sandstone reservoir, accurate permeability curves can be acquired. Unlike traditional methods that rely on empirical constants or unimodal assumptions, this approach provides a physics-based, sample-specific link between NMR log response and the nano-CT scanning data. The resulting model offers a practical and more accurate method for permeability evaluation in challenging tight sandstone reservoirs.