Capillary trapping quantification in sandstones using NMR relaxometry

Abstract

Capillary trapping of a non‐wetting phase arising from two‐phase immiscible flow in sedimentary rocks is critical to many geoscience scenarios, including oil and gas recovery, aquifer recharge and, with increasing interest, carbon sequestration. Here we demonstrate the successful use of low field 1H Nuclear Magnetic Resonance [NMR] to quantify capillary trapping; specifically we use transverse relaxation time [T2] time measurements to measure both residual water [wetting phase] content and the surface‐to‐volume ratio distribution (which is proportional to pore size] of the void space occupied by this residual water. Critically we systematically confirm this relationship between T2 and pore size by quantifying inter‐pore magnetic field gradients due to magnetic susceptibility contrast, and demonstrate that our measurements at all water saturations are unaffected. Diffusion in such field gradients can potentially severely distort the T2‐pore size relationship, rendering it unusable. Measurements are performed for nitrogen injection into a range of water‐saturated sandstone plugs at reservoir conditions. Consistent with a water‐wet system, water was preferentially displaced from larger pores while relatively little change was observed in the water occupying smaller pore spaces. The impact of cyclic wetting/non‐wetting fluid injection was explored and indicated that such a regime increased non‐wetting trapping efficiency by the sequential occupation of the most available larger pores by nitrogen. Finally the replacement of nitrogen by CO2 was considered; this revealed that dissolution of paramagnetic minerals from the sandstone caused by its exposure to carbonic acid reduced the in situ bulk fluid T2 relaxation time on a timescale comparable to our core flooding experiments. The implications of this for the T2‐pore size relationship are discussed.

Capillary Trapping of CO2 in Sandstone Using Low Field NMR Relaxometry

“…A review of published CO 2 core flooding data found about 95% of the data for sandstone were located between the Land model curves for values of C ranging from 0.7 to 2.0 (Krevor et al, ). All the results obtained for the present NMR‐based measurements with CO 2 ‐brine systems and for the previously published N 2 ‐water system (Connolly et al, ) consistently lie within these bounds. Berea sandstone demonstrated the largest trapped volume of CO 2 , which we speculate reflects the fact that it contains the largest pores.…”

“…One alternative approach for studying fluid wettability in rock cores is via nuclear magnetic resonance (NMR) relaxometry measurements which are inherently sensitive to brine-surface interactions (e.g., wetting) and can also provide quantitative saturation information. Furthermore, many other related pore fluid and porous medium properties can be measured including self-diffusion of pore fluids for fluid typing and tortuosity (e.g., Arns et al, 2011;Zecca et al, 2018), NMR relaxation times for saturation as a function of pore size (e.g., Connolly et al, 2017;Kleinberg et al, 1994), and multidimensional relaxometry for pore to pore fluid exchange characterization and surface adsorption energetics (e.g., Callaghan et al, 2007;Robinson et al, 2018).…”

“…Consequently, changes in the measured T 2 probability distributions can be related to pore size occupation by the brine solution during and after core flooding. Results are then compared to previously published NMR results for N 2 capillary trapping in similar sandstones (Connolly et al, 2017).…”

Dielectric Polarization Studies in Partially Saturated Shale Cores

“…If the magnitude of penalty function is too small, then the solution appears under‐regularized resulting in a highly oscillatory and generally unphysical g distribution. Conversely, if the penalty function is too large, then not enough weight is given to the exact analytical solution resulting in very broad, under‐defined g distribution (Connolly et al, ). To select the alpha value, we have used the automated method of generalized cross validation (GCV), which is described in detail elsewhere (Hollingsworth & Johns, ; Wahba, ; Wilson, ).…”

Understanding the microstructural evolution of hypersaline cemented paste backfill with low-field NMR relaxation

“…This robust mathematical inversion technique makes no assumptions about the shape of the resultant probability distributions, and has been successfully implemented across a range of NMR data analysis protocols, including relaxation time distributions in porous media (e.g. Connolly et al, 2017;Fridjonsson et al, 2013;Robinson et al, 2017) and droplet sizing methods (Hollingsworth and Johns, 2003). The inversion algorithm used in this work was written in MATLAB (MathWorks Inc.) and first used by Griffith et al.…”