Permeability determination from NMR logs is of particular interest in unconventional reservoir characterization. Permeability logs have direct impact on calculating producibility and recoverability of hydrocarbons, and are used to optimize horizontal-well placement for hydraulic fracturing and production.
Nevertheless, both the Timur-Coates and SDR permeability models fail in unconventional shale formations such as shale-gas and shale-oil, because these models do not explicitly account for the tortuosity of the porous medium, where tortuosity is typically large in unconventional organic-shale since porosity is typically low.
A new method is developed to estimate permeability through the combination of porosity measurements, tortuosity estimation and MICP pore-throat size measurements. Water-saturated cores are submerged in D2O (i.e. heavy water) brine and the decay in NMR porosity with time is monitored to determine the tortuosity of the cores. The NMR derived tortuosity is used, together with the MICP pore-throat size, to predict the matrix permeability using a modified Carman-Kozeny permeability equation.
In this report, cores from the unconventional Point-Pleasant formation with matrix permeabilities (GRI) in the range of 1 nD – 30 nD, and movable-fluid porosities (NMR T2) in the range of 3 pu – 7 pu are examined. There is a good agreement between our modified Carman-Kozeny estimated permeability and measured permeability. This new NMR core-analysis procedure can also be used to calibrate permeability from NMR logs.
This new permeability method for unconventional formations, correlates permeability with pore-throat size, tortuosity and porosity using the modified Carman-Kozeny equation. It predicts permeability without any adjustable parameters, which is an improvement compared with two well-known NMR permeability models, the Timur-Coates and SDR models.
It is well known that pore-throat size and permeability are strongly correlated, therefore a measure of pore-throat size can in principle be used to estimate permeability through empirical models. For certain sedimentary rocks, the distribution in NMR transverse relaxation time (T2) is a good measure of the pore-body size distribution. These conjectures form the basis of two well-known NMR permeability models: (1) the Timur-Coates model [Timur 1969, Coates 1991], and (2) the SDR (Schlumberger-Doll Research) model (a.k.a. the Kenyon model) [Kenyon 1988]. Both NMR permeability models are based on the underlying assumption that the pore-body to pore-throat ratio (BTR) is roughly constant for all pores, and therefore the T2 distribution is a good measure of the pore-throat distribution. Both models were originally developed for conventional sandstones, though both have also been used for conventional carbonates with variable success. More recently, machine learning techniques have been used to calibrate the Timur-Coates and Kenyon permeability models on conventional logs using core permeability data [Kausik 2020].