Effect of CT image size and resolution on the accuracy of rock property estimates. Issue 5 (29th May 2017)

• Record Type: Journal Article
• Authors: Bazaikin, Y.; Gurevich, B.; Iglauer, S.; Khachkova, T.; Kolyukhin, D.; Lebedev, M.; Lisitsa, V.; Reshetova, G.
• Is Part Of: Journal of geophysical research. Volume 122:Issue 5(2017)
• Page Start: 3635
• Page End: 3647
• Publication Date: 2017-05-29
• Subjects: statistically representative volume -- transport properties of the rock -- topology and geometry of the pore space; Geomagnetism -- Periodicals; Geochemistry -- Periodicals; Geophysics -- Periodicals; Earth sciences -- Periodicals; 551.1
• DOI: 10.1002/2016JB013575
• Languages: English
• ISSNs: 2169-9313

Abstract

Abstract: In order to study the effect of the micro‐CT scan resolution and size on the accuracy of upscaled digital rock property estimation of core samples Bentheimer sandstone images with the resolution varying from 0.9 μm to 24 μm are used. We statistically show that the correlation length of the pore‐to‐matrix distribution can be reliably determined for the images with the resolution finer than 9 voxels per correlation length and the representative volume for this property is about 15 3 correlation length. Similar resolution values for the statistically representative volume are also valid for the estimation of the total porosity, specific surface area, mean curvature, and topology of the pore space. Only the total porosity and the number of isolated pores are stably recovered, whereas geometry and the topological measures of the pore space are strongly affected by the resolution change. We also simulate fluid flow in the pore space and estimate permeability and tortuosity of the sample. The results demonstrate that the representative volume for the transport property calculation should be greater than 50 correlation lengths of pore‐to‐matrix distribution. On the other hand, permeability estimation based on the statistical analysis of equivalent realizations shows some weak influence of the resolution on the transport properties. The reason for this might be that the characteristic scale of the particular physical processes may affect the result stronger than the model (image) scale. Key Points: Correlation length of pore‐to‐matrix distribution is statistically reliably determined for resolution finer than 9 voxels per correlation length The statistically representative volume for pore‐to‐matrix distribution is about 15 correlation lengths The representative volume for the transport property calculation should be greater than 50 correlation lengths of pore‐to‐core distribution