Robust determination of surface relaxivity from nuclear magnetic resonance DT2 measurements

Nuclear magnetic resonance (NMR) is a powerful tool to probe into geological materials such as hydrocarbon reservoir rocks and groundwater aquifers. It is unique in its ability to obtain in situ the fluid type and the pore size distributions (PSD). The T-1 and T-2 relaxation times are closely related to the pore geometry through the parameter called surface relaxivity. This parameter is critical for converting the relaxation time distribution into the PSD and so is key to accurately predicting permeability. The conventional way to determine the surface relaxivity rho(2) had required independent laboratory measurements of the pore size. Recently Zielinski et al. proposed a restricted diffusion model to extract the surface relaxivity from the NMR diffusion-T-2 relaxation (DT2) measurement. Although this method significantly improved the ability to directly extract surface relaxivity from a pure NMR measurement, there are inconsistencies with their model and it relies on a number of preset parameters. Here we propose an improved signal model to incorporate a scalable L-T and extend their method to extract the surface relaxivity based on analyzing multiple DT2 maps with varied diffusion observation time. With multiple diffusion observation times, the apparent diffusion coefficient correctly describes the restricted diffusion behavior in samples with wide PSDs, and the new method does not require predetermined parameters, such as the bulk diffusion coefficient and tortuosity. Laboratory experiments on glass beads packs with the beads diameter ranging from 50 mu m to 500 mu m are used to validate the new method. The extracted diffusion parameters are consistent with their known values and the determined surface relaxivity rho 2 agrees with the expected value within +/- 7%. This method is further successfully applied on a Berea sandstone core and yields surface relaxivity rho(2) consistent with the literature. (C) 2015 Elsevier Inc. All rights reserved.