APLIN, ANDREW C., YUNLAI YANG and STEVE R. LARTER
NRG, University of Newcastle upon Tyne, UK

Abstract: Rapid, Wireline Log Assessment Of Shale Properties For Basin Modelling

Accurate predictions from basin models necessitate accurate input data. For fluid flow and pressure, a critical area of uncertainty is the input data for shales: their compressibility and porosity - permeability relationships. Using a large (200+ samples) database of well characterised mudstones, we show here that both the compressibility (porosity effective stress) and porosity - permeability relationships of shales can be predicted quite confidently from their clay fraction (% <2 micro meters particles). Micro to sub-nanoDarcy permeabilities are predicted to a factor of +/- 3 as a dual function of clay fraction and porosity. When constructing basin models, it is impractical to routinely measure shale properties. We therefore show here how the compressibility and porosity - permeability relationships of shales can be assessed from wireline log data. We have trained Artificial Neural Networks (ANNs) to predict the clay fraction and grain density (and thus porosity) of shales from a standard log suite. The ANNs trained well and quite accurately predicted clay fraction and porosity in areas outside those in which they were trained. Clay fraction and porosity then allow one to predict shale permeability, compressibility and porosity - permeability behaviour, on a metre scale. These new techniques raise the possibility of defining, for the first time, the 3D anatomy and fluid- flow structure of large shale bodies.

AAPG Search and Discovery Article #90928©1999 AAPG Annual Convention, San Antonio, Texas