Fracture Identification in Image Logs Using Image Processing Techniques and Algorithm Genetic

In naturally fractured reservoirs, fractures play a main role in production, and fracture identification is very important in reservoir development and management. Borehole image log, which is a high resolution “pseudo picture” of the borehole wall, is a powerful tool for fracture study. These logs provide critical information about the orientation, depth, and type of natural fractures. Currently, there is no comprehensive algorithm for the automatic identification of fracture parameters in image logs, and the interpretation of these logs is often done manually. This process might become erroneous when the interpreter is less experienced. The present study uses image analysis and processing techniques, as well as genetic algorithms to detect fractures in image logs automatically. In this method, the points related to fractures are first extracted from the image by a classification method. Then, the number, depth, dip, and dip direction of fractures are determined on the extracted points by using genetic algorithm. This method is performed on a part of two image logs (4 and 8 pads) of two wells located in two oilfields in the south of Iran. Despite the sensitivity of the proposed method to the noises of the image, it successfully estimated the number, dip, and dip direction of fractures for both studied wells with an accuracy of 70%.