A Laboratory Study to Determine the Effect of Field Strength and Magnetic Susceptibility on the NMR Estimated Water Content in Unconsolidated Sediments

Abstract:

Geophysical nuclear magnetic resonance (NMR) well logging data can provide direct information about subsurface water content. While NMR water content estimates are known to be accurate in low magnetic susceptibility materials, it has often been assumed that NMR measurements cannot be used in high magnetic susceptibility materials due to internal magnetic field inhomogeneities that arise due to magnetic susceptibility contrasts in the material. In this study we compare the NMR estimated water content using laboratory measurements made at two low magnetic field strengths (with Larmor frequencies of 275 kHz and 2 MHz) on both synthetic and natural unconsolidated sediments with a range of magnetic susceptibility values.

NMR measurements were collected on seven water-saturated materials with magnetic susceptibility values spanning three orders of magnitude (3.6x10^-6 SI to 7.0 x10^-3 SI). T₂ relaxation time data was collected with echo times, t_E, ranging from 200 to 3000 μs. The results show that for the materials with low magnetic susceptibilities (< 5x10^-4 SI), the total water content was accurately estimated at both field strengths. For the materials with high magnetic susceptibilities (> 5x10^-4 SI) the water content was more accurately estimated using the data collected at 275 kHz (> 80% detected at t_E = 400 μs) than the data collected at 2 MHz (< 40% detected at t_E = 400 μs). Furthermore, the 275 kHz data showed water content underestimation errors increased only slightly with increased t_E, compared to substantial increases in errors for the 2 MHz data as t_E was increased. This finding suggests that there is an advantage for collecting measurements at lower field strengths even for long t_E. We explain the differences in the water content estimates at the two field strengths by considering the shape of the echoes and the coil and pulse bandwidths, and find excellent agreement with the range of collected NMR data.