Effects on Digital Rock Physics Models from Variable Computed Tomography Scans Settings

Description

The present contribution aims to understand the effect of Computed Tomography (CT) acquisition settings on digital rock physics models. We conducted 32 scans of 4 Berea Sandstone samples. In order to understand the effects of setup parameters on the calculated density of the samples, each scan had different set up conditions. We also varied post acquisition filtering, including applying different beam hardening corrections, and various noise filters. We observe that by using phantoms of known density scanned alongside our samples, we can calibrate CT attenuation to density in most circumstances. X-ray source current, numerical noise filtering, exposure time, number of projections, and numerical bit size seem to have a little effect on the final density estimation. Scans with very low source voltages (<80 kV), as well as some with physical copper filtering over the X-ray source, may not be effective for physical property estimations. We also found that density estimations are only accurate if a specific, and unique, beam hardening correction is applied. We observed that when the beam hardening correction is properly selected, the R2 of the CT-number to density calibration curve is above 0.999 and density estimates are accurate. Scans that cannot be used for accurate property estimations can be easily identified using this metric.