Automatic Identification of Fractures Using a Density-Based Clustering Algorithm with Time-Spatial Constraints

In shale gas hydraulic fracture monitoring or rock acoustic emission experiments, fracture plane identification is always a complex task. Conventional approaches typically use the source locating results derived from the micro-seismic event and then interpret the fracture plane using manual qualitative analysis. Large errors typically occur due to manual operations. On the other hand, the density-based clustering algorithm with spatial constraints is widely used in geographic information science, biological cells science and astronomy. It is an automated algorithm and can achieve good classification results. In this paper, we introduced the above-mentioned clustering algorithm with spatial constraints to fracture identification applications. Moreover, because micro-seismic events are 4D in nature, every micro-seismic event has both time and space information. Hence, we improve the conventional clustering algorithm by incorporating a time constraint. We test the proposed method using rock acoustic emissions data and compare our fracture identification results with CT scan images; the comparison clearly shows the effectiveness of the proposed method.