Relative Permeability Measurement in Rock Fractures

The petroleum industry always requires precise measurement of relative permeability. When it comes to the fractures, the measurement becomes extremely difficult since the saturation in the fractures is hard to calculate. In this study, we are aiming to provide a machine learning approach to distinguish voxel-based water and gas in rock fractures based on Computed Tomography. This research is divided into two parts: the first part is to extract fracture features from a set of noisy CT scanning data; the second part involves water and air classification. We use a machine learning assisted boosted depth first search. We use the assumption that fractures would be continuous or at least piecewise continuous to devise this algorithm. For our project, we seek to find f(N(x)) to {air, water, rock} which optimizes the mapping of the pixel to the correct class. Two-phase data will be used to test the machine learning results and to provide a distribution of water and gas in fractures. Finally, the fracture has been successfully extracted and the relative permeability curve for fractures have been obtained based on our experimental data. This could be a significant contribution to shale gas production, enhanced oil recovery and hydraulic fracturing activities.