The Application of Artificial Digital Models in X-Ray Computed Tomography (CT) of the Core in Solving the Problem of Binarization of the Void Space of Reservoir Rocks

The X-ray tomography method has several advantages, including its non-destructiveness and the ability to visualize the rock skeleton and pore space in three dimensions. However, one of the main challenges of this method is the limited resolution when studying core samples that are 30 millimeters in diameter, which is typical for petrophysical analysis. In these samples, a significant portion of pores have dimensions smaller than the resolution capabilities of most X-ray tomographic systems, making it impossible to accurately determine the boundary between the pore and skeleton structures in tomograms, nor visualize the entire pore volume. To verify this hypothesis, tomograms from real oil and gas samples were analyzed. The resulting histograms of X-ray densities revealed that it is not possible to directly measure the threshold value of X-ray density that defines the “skeleton-pore” boundary. In order to solve the problem of estimating boundary values, a technique is proposed in this work that suggests using artificial digital models – phantoms. This approach has been previously used mainly in computer modeling, but it has not been used much in petroleum geology. The main advantage of using phantoms is complete control over the set pore space parameters and X-ray density of the skeleton, which cannot be achieved on real samples. A computational experiment was conducted in the work, where 124 core phantoms with specific porosity characteristics were generated using numerical modeling. These phantoms were then converted into tomograms, allowing us to determine statistical characteristics of the values for X-ray densities of the samples at the reconstruction stage. Based on the statistical analysis of the X-ray density distribution in the sample, we determined the boundary values that are most suitable for reliable void space detection. Using regression and correlation methods, we developed a model to estimate the optimal boundary value for X-ray density in void space allocation. We proposed an algorithm for determining and applying this value in the analysis of core X-ray CT data.This model was tested on real samples that were not used in the development of the forecast model. The use of the proposed model for predicting boundary values on obtained tomograms demonstrated a high degree of consistency with actual data.