[M-GI29] Data-driven geosciences

[MGI29-P04] A New Descriptor of Flow-Channel Structures Based on Topological Data Analysis

*Togo Hasumi¹, Keiichiro Goto¹, Marco Fuchs², Philipp Blum², Anna Suzuki¹ (1.TOHOKU UNIVERSITY, 2.Karlsruhe Institute of Technology)

Keywords:Yagiyama

Fracture surfaces in rocks have heterogeneous aperture distributions due to contact and separation between fractures, and preferential flow channels are formed due to these heterogeneous apertures. Only the structures of the preferential flow-channel, which is connected from inlet to outlet in the rock structure, affects the flow characteristics. If the connectivity information of flow channels can be extracted, it is expected that the huge amount of data on the structure can be reduced to the information necessary to evaluate the flow characteristics. Topological data analysis has been focused as a method to evaluate connectivity from images. In a previous study, we showed that the topological data analysis can be applied to images of fracture network structures to extract important structures in preferential flow channels and to estimate the flow characteristics (i.e., permeability). The objective of this study is to extract structures of preferential flow-channels from image information of a single crack with heterogeneous aperture distribution and to analyze flow properties. From a voxel image of rock samples (8.2 mm x 116.6 mm x 392.4 mm) with a single fracture, the channel structure was extracted by topological data analysis and the permeability rate was estimated. The estimated permeability was consistent with the experimental results. This method may enable dimensional reduction of rock structures by describing a new descriptor of rock flow channels and would facilitate expansion to computationally expensive multiphysics problems and upscaling problems that require a wider area to be covered.