Numerical prediction of carbonate elastic properties based on multi-scale imaging

Elastic properties predictions of rocks using numerical simulations are generally overestimated compared to laboratory measurements regardless of the algorithms used. This overestimation is prevalent among sandstones as well as carbonate rock types but the degree of the mismatch between the two results is much higher for carbonates due to the complex pore structures and heterogeneity at the pore scales. One key reason attributed towards the systematic overestimation is imaging system’s limitation to resolve pore structures below its threshold resolution at representative volumes. To study the effect of this limitation, we developed a multi-scale imaging approach and “up-scaling” framework to improve the numerical predictions of the linear, isotropic elastic properties of a standard dolomite rock using the Digital Rock Physics approach. We defined up-scaling as the process of integrating information from high resolution images (obtained at micro scale) to improve prediction using the lower resolution images obtained at full-plug scale covering a larger representative volume. A combination of multi-resolution (40, 13, 5 and 1μm) X-ray micro computer tomography and Focus Ion Beam combined with Scanning Electron Microscope (FIB/SEM) images for the dolomite rock were then utilized. We compared numerically simulated linear elastic and isotropic moduli to in-house laboratory acoustic velocity test results performed on the same dolomite carbonate sample that was used for imaging. Results showed a reduction of the overestimation from 8.9% to 4.5% for predicted P-wave velocity and from 11.9% to 7.8% for predicted S-wave velocity when the multi-scale imaging approach was used. •Spatially registered multi-scale imaging helps bridge the volume vs imaging resolution dilemma.•Integration of multi-scale information can improve numerical elastic properties prediction.•Simulations and laboratory results compared for same carbonate plug material and samples to reduce uncertainties.