Electrical formation factor of clean sand from laboratory measurements and digital rock physics
Electrical properties of rocks are important parameters for well-log and reservoir interpretation. Laboratory measurements of such properties are time-consuming, difficult, and impossible in some cases. Being able to compute them from 3-D images of small samples will allow for the generation of a massive amount of data in a short time, opening new avenues in applied and fundamental science. To become a reliable method, the accuracy of this technology needs to be tested. In this study, we developed a comprehensive and robust workflow with clean sand from two beaches. Electrical conductivities at 1 kHz were first carefully measured in the laboratory. A range of porosities spanning from a minimum of 0.26–0.33 to a maximum of 0.39–0.44, depending on the samples, was obtained. Such a range was achieved by compacting the samples in a way that reproduces the natural packing of sand. Characteristic electrical formation factor versus porosity relationships were then obtained for each sand type. 3-D microcomputed tomography images of each sand sample from the experimental sand pack were acquired at different resolutions. Image processing was done using a global thresholding method and up to 96 subsamples of sizes from 2003 to 7003 voxels. After segmentation, the images were used to compute the effective electrical conductivity of the sub-cubes using finite-element electrostatic modelling. For the samples, a good agreement between laboratory measurements and computation from digital cores was found if a sub-cube size representative elemental volume (REV) was reached that is between 1300 and 1820 µm3, which, with an average grain size of 160 µm, is between 8 and 11 grains. Computed digital rock images of the clean sands have opened a way forward for obtaining the formation factor within the shortest possible time; laboratory calculations take 5 to 35 d as in the case of clean and shaly sands, respectively, whereas digital rock physics computation takes just 3 to 5 h.