Calculating Effective Elastic Properties of Berea Sandstone Using the Segmentation-Less Method Without Targets

We propose a new method to compute elastic properties of rocks from computed tomography (CT) images. A CT volume captures X-ray attenuation, which is scaled in a CT number unit. Our method improves an existing segmentation-less method where the CT volume is directly converted to elastic property arrays without using segmentation. A drawback of the existing segmentation-less method is that it typically requires to scan physical targets of known density along with the sample. Targets map a CT volume into petrophysical property arrays by comparing the CT numbers of the sample with those of the targets. We introduce a segmentation-less workflow that no longer requires targets but uses pseudotargets that are defined within the sample. Pseudotargets are composed of extrema CT numbers in the CT volume. We assume that regions containing extrema CT numbers represent pristine materials. The average CT numbers of such regions are paired to the pseudotarget densities to create a CT number-to-density conversion function. After density-to-porosity and porosity-to-elastic properties conversion, we simulate wave propagation on two Berea sandstone cores to estimate effective P and S wave velocities. To assess uncertainty, computed velocities are compared to laboratory measurements. We benchmark our technique against a segmentation-less method, as well as the more commonly used segmentation-based method. Among the three methods, the segmentation-less method without targets yields the least absolute velocity error of 4.5%, whereas segmentation and segmentation-less method yield +29% and +4.7% error, respectively. We conclude that the segmentation-less method without targets can efficiently compute elastic properties of rocks from CT volumes, serving as a technique to create better subsurface models.