Improved estimates of formation factor for combined electrical and nuclear magnetic resonance models of permeability of sandstone cores

Abstract:

In spite of the importance of permeability in controlling numerous hydrogeological and biogeochemical processes, the property can be exceptionally difficult to measure directly in the field. Recently, nuclear magnetic resonance (NMR) has become an increasingly popular method, both in the lab and the field, for hydrogeophysical investigations due to its sensitivity to water content and pore surface area. Additionally, previous work has shown that the electrical formation factor can be used as a proxy for the tortuosity of the pore space—a parameter NMR is incapable of detecting—in permeability models. However, the formation factor is impossible to accurately measure in the field using DC electrical methods, as the measured conductivity cannot be decomposed into the fluid and surface conduction components. Therefore, our approach is to use induced polarization (IP) and spectral induced polarization (SIP) in the laboratory to correct for the influence of surface conductivity in the formation factor calculation. The corrected formation factor can then be used along with NMR parameters for more accurate permeability estimation.

Laboratory SIP and NMR datasets were acquired on 40 sandstone cores with a range of permeabilities spanning six orders of magnitude as estimated from gas permeameter measurements. We examine how different estimates of the electrical formation factor can be combined with the NMR transverse relaxation time to estimate permeability. Specifically, we compare the electrical formation factor measured at high and low pore-fluid salinity with the formation factor derived using IP and SIP. Using both empirical and mechanistic petrophysical relationships, we explore the utility of IP- and SIP-corrected formation factors in tandem with NMR parameters for permeability prediction as compared to the low-salinity formation factor typically measured in the field. Furthermore, we develop our models using IP and SIP data that may be acquired in the field. Going forward, our goal is to develop robust electrical-NMR models that can be applied to laboratory, well-log and surface geophysical data.