The petrophysical properties of shale gas reservoirs

A significant challenge to the petrophysical evaluation of shale gas systems can be attributed to the conductivity behaviour of clay minerals. This is compounded by centimetre to sub-millimetre vertical and lateral heterogeneity in formation geological and therefore petrophysical properties. Despite this however, we remain reliant on Archie based methods for determining water saturation (Sw), and hence the free gas saturation (1-Sg) in shale gas systems. There is however significant uncertainty in both how resistivity methods are applied and the saturation estimates they produce, due largely as Archie parameter inputs (e.g. a, m, n, and Rw) are difficult to determine in shale gas systems, where obtaining a water sample, or carrying out laboratory experiments on recovered core is often technically impractical. This research assesses the geological implications for, and controls on, variations in pseudo Archie parameters in the Bossier and Haynesville Shale Formations in the northern Gulf of Mexico basin. Investigation has particularly focused on the numerical analysis and systematic modification of Archie parameter values to minimise the error between core SW (Dean Stark analysis) and computed Sw values. Results show that the use of optimised Archie parameters can be effective in predicting SW, particularly in the Haynesville formation, but identifies systematic bias in generated Archie parameters that precludes their accurate physical interpretation. Analysis also suggests that variability in the resistivity (Rt) log response is the principal source of error in Sw estimates in the Bossier Shale. Moreover, results suggest that where clay volume exceeds 28%, the resistivity response becomes increasingly variable and elevated, indicating an apparent clay associated ‘excess resistivity’. This is explained by a geologically consistent model that links increasing clay volume to bulk pore water freshening, supported by empirical adaptations that allow for improved Archie parameter selection and a further reduction in the error of Sw estimates.