Directional- and scale-dependent permeability estimations from CT images

While permeability is one of the most important parameters for describing porous media such as subsurface reservoirs, it is also one of the most challenging parameters to obtain. The challenge arises from a lack of methods, e.g., down-hole logging tools, to directly measure permeability. Another challenge is that the permeability is a transport property, and therefore dependent on scale and direction. Whole core CT images of core material unlock new possibilities to investigate permeability by providing high-resolution three-dimensional structural information. Current whole-core CT imaging provides images with millimeter resolution, thereby resolving lamina scale heterogeneities in the core material known to affect the anisotropy of the permeability. In this article, we will use a 2D cross-section of a whole core CT scan to investigate how small-scale heterogeneities affect permeability. For this comparison, we will use different upscaling techniques and optimization to reveal the relation between distribution in CT values and permeability. We observe that a linear relation between CT values and permeability at the voxel scale gives a fair estimate of the permeability at the scale of core plugs. We also observe that computationally light averaging techniques provide similar results as more computationally heavy numerical solutions. For our data, the main controlling factor for permeability at a centimeter scale is the average CT value, while the fine-scale heterogeneities have a clear influence on the scale dependence of the permeability. The directional dependence of the permeability can to some extent be derived from the observed fine-scale heterogeneity in the CT images, and the distribution asserts a larger influence on permeability anisotropy at coarser scales. This work displays how CT images can serve as a foundation for handling the scale dependence of permeability. • This paper presents a method to infer pixel-based permeability from CT images. • It correlates measured core plug permeabilities with upscaled CT-based permeabilities • Pixel-based permeability enables the calculation of permeability at different scales. • An output from image-based permeability is continuous permeability logs. • The method has been applied to an imaged well. • The application reveals a clear influence of scale. • The influence of upscaling methods is limited. • The method presents a pathway for obtaining reservoir-model scale permeability data.