A laboratory study to determine the effect of pore size, surface relaxivity, and saturation on NMR T₂ relaxation measurements

Abstract

In this study, we present laboratory experiments investigating the effect of pore size and surface relaxivity ρ2 on the nuclear magnetic resonance response of variably saturated sands that relax both within and outside the fast diffusion regime. We measured the NMR response of sands with a range of grain sizes (129 to 753 μm), which resulted in samples with different pore sizes, and a range of iron concentrations (0.07% to 0.38%), which resulted in sands with different ρ2 values. The laboratory results showed that the relation between relaxation time and water saturation depended on the regime in which relaxation occurred. For samples relaxing in the fast diffusion regime (i.e., small pores and low surface relaxivity), the relation between relaxation time and water saturation was linear; for the remaining samples, the relation between relaxation time and saturation demonstrated a power‐law relationship with an exponent greater than one. In addition, we performed numerical simulations based on common pore‐filling mechanisms (i.e., capillary tubes or thin films). The numerical simulations did not predict the experimental results for the relative surface relaxation versus saturation trends. We conclude that, in addition to the diffusion regime in which relaxation occurs, the shape of the relaxation time versus saturation curve depends on the pore‐filling mechanism, the value of ρ2, and the pore size. The dependence of NMR relaxation on the pore‐filling mechanism of saturating porous media may complicate efforts to develop a relation between relaxation time and saturation for use in characterizing unsaturated porous media.

The NMR Relaxation Response of Unconsolidated Sediments during Drainage and Imbibition

“…Consistent with previous studies (e.g., Keating and Falzone, 2013), samples with smaller d mean and/or larger ρ 2 have distributions centered at shorter relaxation times; samples with larger d mean and/or smaller ρ 2 have distributions centered at longer relaxation times. Also consistent with previous studies, the location of the T 2 distribution shifted with saturation (Bird and Preston, 2005; Falzone and Keating, 2016; Ioannidis et al, 2006; Mohnke, 2014). During drainage (Fig.…”

“…The NMR parameters determined from the saturated samples, T 2ML −1 , T 2D −1 , T 2S −1 , and ρ 2, are shown in Table 3 and were calculated using measured values of T 2B −1 , which varied between 0.383 and 0.407 s −1 , consistent with previous measurements on deionized water (Falzone and Keating, 2016). For the synthetic sands and for StH, T 2ML −1 showed no dependence on t E 2 indicating T 2D −1 ~ 0 for these samples; for StL, T 2ML −1 showed a small dependence on t E 2 .…”

“…Numerous studies have shown that the NMR relaxation time distribution changes with saturation (e.g., Bird and Preston, 2005; Costabel and Yaramanci, 2011b; Falzone and Keating, 2016; Ioannidis et al, 2006; Jaeger et al, 2009; Mohnke, 2014). Unlike in saturated systems, where the NMR response is a function of the pore size distribution of the measured sample, in unsaturated systems the NMR response is a function of the water‐filled pore space and thus is related to the water‐filled pore size distribution.…”

Characterizing the Critical Zone Using Borehole and Surface Nuclear Magnetic Resonance

“…The short relaxation times in the vadose zone are thought to be caused by a preference for filling the smallest pores first, which usually have shorter relaxation times (Walsh et al, 2014). Furthermore, the relaxation times can change as a function of saturation and have been shown to exhibit hysteretic behavior (Falzone and Keating, 2016a, 2016b).…”

“…Nuclear magnetic resonance is favorable for characterizing groundwater because the magnitude of the signal is dependent on the volume of groundwater and, under most geologic conditions, the signal characteristics can be related to the surface/pore volume ratio (Gallegos et al, 1988; Mohnke and Yaramanci, 2008; Keating and Falzone, 2013; Falzone and Keating, 2016a), which has been shown to be related to hydraulic conductivity (Timur, 1969; Kenyon, 1997; Walsh, 2008; Weller et al, 2010). The use of surface NMR to characterize fractured rock systems is uncommon due to (i) the low water content of unweathered crystalline rock (average matrix porosities less than ∼5%; Begonha and Braga, 2002; Goodfellow et al, 2016), (ii) the presumed presence of magnetic gradients caused by Fe minerals commonly present in igneous rocks, which are expected to distort the surface NMR signal (e.g., Grunewald and Knight, 2011), and (iii) lateral heterogeneities of fracture networks, which break the common layered assumption used in surface NMR forward modeling (Legchenko et al, 2006; Müller‐Petke et al, 2016).…”