Statistical Analysis of Well Fracture Data to Distinguish and Extract Fracture Corridors from Diffuse Fractures

Recognizing fracture swarms from diffuse fractures, and interpreting them as being related to corridors, thin rock beds or facies, is a critical and often time-consuming step in naturally fractured reservoir (NFR) characterization. On the one hand, it is necessary to decide whether or not visually aggregated fractures are indeed anomalous and should be considered as belonging to a single fracture corridor. On the other hand, user-friendly tools are needed to differentiate fractures within and outside corridors so as to properly take them into account in NFR models. An innovative statistical approach, associated with post-processing tools for automatic and assisted recognition of within-corridor fractures, is presented. Based on the SiZer method (SIgnificant ZERo crossings of derivatives), the statistical identification of fracture swarms relies on statistical hypothesis testing to detect significant fracture density trends along wells at different observation scales. Knowing zones of very likely increasing, stationary or decreasing fracture densities, a post-processing algorithm allows to distinguish fractures that are inside and outside of identified corridors, to mark them distinctively and to generate equivalent data that can be used to characterize and model large scale fracture corridors. The paper presents the approach and illustrates its performances on a real case study.