Nmr Surface Relaxivity and Diffusion Effects in Grain Packs

The surface relaxivity ρ is an important and fundamental parameter for the interpretation of T2 distributions. To use laboratory data in the field, one must know the various effects that will influence ρ, such as temperature and mineralogy. Recent works suggest that ρ value is fluctuating more than expected and yields errors in water saturation estimation. We studied low field (2.2 MHz proton) nuclear relaxation in calibrated silicon carbide (SiC) grain packs in a large range of grain size (8-150 μm) to avoid ambiguity due to pore size distribution of most consolidated reservoir rocks. To isolate the origins of surface relaxation, we varied the temperature from 85°C to 34°C. Measurements were performed with both high and low amount of paramagnetic impurities at the surface of the grains When the amount of paramagnetic impurities is low, the surface relaxation is in the range of natural rocks and we found a linear relationship between T1 and T2 and pore size, indicating a relaxation process limited by the surface. When the temperature is increased, the linear relationship is still valid but there is a significant decrease of the relaxation times. Similar results obtained on porous silica glasses suggest that this unexpected phenomenon is due to a surface diffusion of proton species. Introduction Low field nuclear magnetic resonance is now routinely used to determine important reservoir properties such as porosity, water saturation and permeability. However, they are still important difficulties related to the determination of surface relaxivity. This fundamental parameter, linking the measured relaxation time T2 to the surface-to-volume ratio (S/V) of the rock's porous space, can be determined by various methods and varies accordingly by several orders of magnitude. The most common method used is the comparison of T2 distribution with mercury injection capillary pressure curves (MICP) because these measurements are simple and performed routinely for other purposes such as rock typing. However, the underlying strong assumption of a correlation between pore throats and pore bodies is only valid for 'simple' rocks for which conventional logging data may suffice. Therefore, such an approach cannot be generalized. Recent works confirm this expected weakness and indicate challenging questions about surface relaxivity; for sandstones, an extensive experimental study shows that the link between T2 and MICP is unpredictable (Lowden et al., 1998); for carbonates, Boyd and El Amam (1998) observed that the cut-off value used to calculate water saturation was not only higher than expected but varies by a factor of four for the samples studied. This implies that NMR logs can only be properly calibrated where cores have been taken for laboratory measurements and that the extrapolation of these calibrations to other intervals will be questionable, despite the various improvements of the logging tools.