Insights on formation damage associated with hydraulic fracturing using image analysis and machine learning

Abstract

AbstractHydraulic fracturing is a well‐recognized technique used to produce as much as possible from unconventional resources. Yet, the communication between the main fracture and the fracture process zone (FPZ) around the primary fracture is still a mystery. Some research was conducted on actual samples to statistically examine the behaviour of these parameters along and parallel to the main fracture and the degree of damage/enhancement induced in the FPZ. Therefore, to understand the formation damage near the hydraulic fracture and its relation to the induced fractures, I am introducing a comprehensive analysis to shed light on some of the observations that I drew through the investigation throughout the previous years. Image analysis was performed on tight core samples under the triaxial compression test in addition to other sources of data and image analysis using data mining and machine learning. The number, length, and aspect ratios of the microcracks were measured. The drainage distance from the main fracture and the development of the microcracks were analyzed. A periodic cycle distribution was apparent around the main fracture and was related to the orientation of the microcracks and the surface roughness of the main hydraulic fracture. These results will help differentiate between the natural and induced microcracks. Additionally, a clear understanding of the relation between the primary fracture and the microcracks, the direction of the microcracks relative to the primary fracture, and the effectiveness of hydraulic fractures in the real world is presented. This work helps find the relation between the natural fractures and the induced fractures in a more realistic way, which helps in the design of the hydraulic fracture implementation and production increase.