Automated Fracture Detection on High Resolution Resistivity Borehole Imagery

Abstract

Abstract This paper proposes a method to detect fractures on High Resolution Resistivity Borehole Imagery. The method provides with quantitative parameters, such as length and contrast of the fractures that can be processed with conventional logs to help characterize reservoir nature and behavior. Thanks to the auto-optimization of the algorithm, this method is easy to operate, does not require expertise and can be used in drilling operation, on the site, in the meantime between log acquisition and casing perforation. The method takes advantage of the fact that:–fractures are planar features secant to the stratification planes and exhibit high resistivity contrast.–stratification planes represents the dominant orientation of planar features that are seen on HR resistivity borehole imagery and, based on this characteristic, they can be detected using statistical analysis of the full image (The method is described on SPWLA logging Symposium transactions 1997, paper O). After detection of the stratification planes, images are processed to filter out the stratification planes and normalize the background, thus making more visible the fracture planes. Fractures segments are detected on each "pad" images using "depth-first search" algorithm to find "min-cost path" and then selected according to length, contrast, linearity and visibility criteria. An image displaying the fracture segments and the associated quantitative parameters is produced. Detailed field examples are used to illustrate the results obtained in different types of lithology. P. 777