LAUSR

Abstract Nuclear magnetic resonance (NMR) relaxometry is a common technique for the petrophysical characterization of sedimentary rocks. The standard interpretation of NMR transverse relaxation responses is based on the assumptions of the fast diffusion limit, dominant surface relaxation, and weak coupling between pores, allowing an interpretation of the NMR response as pore size distribution. In general, three asymptotic relaxation regimes can be defined by comparing dephasing, structural and diffusion length scales. The shortest length scale defines the dominating relaxation regime, which is either the free diffusion, motional averaging, or localization regime. Complex mineralogy and morphology of rocks, especially due to clay minerals, may lead to the co-existence and coupling of multiple relaxation regimes in the presence of surface relaxivity heterogeneity and internal gradient effects. In this work the relationship between morphology of rocks, strength of applied magnetic field, resulting relaxation regimes and observed NMR relaxation responses is studied numerically. Limitations in the discretization of clays typical for micro-CT images are addressed by introducing a fine-scale kaolinite model based on SEM images. The model is utilized to evaluate the field-dependent mean transverse relaxation time and effective diffusion coefficient of clay. These effective parameters are incorporated into random-walk simulations on micro-CT images. Relaxation regimes are analysed spatially for three different types of sandstone as function of external magnetic field strength, CPMG echo time and clay susceptibility. Increase of paramagnetic clay content, field strength and echo-time increases the fraction of spins relaxing in the localization regime within macro-pores, typically being localised in the regions neighbouring clay pockets and grain surfaces. We demonstrate that a significant contribution of the localisation regime to relaxation may lead to incorrect predictions of pore-size distributions and fluid typing. The prefactor of permeability correlations utilizing the logarithmic mean relaxation time varies significantly due to clay effects. Accounting for these effects should enable more robust petrophysical interpretations.