سال انتشار: ۱۳۹۱

محل انتشار: اولین کنفرانس بین المللی نفت، گاز، پتروشیمی و نیروگاهی

تعداد صفحات: ۶

نویسنده(ها):

Mahmoud Seifallahi – School of Electrical and Robotic Engineering, Shahrood University of Technology, Shahrood, Iran
Behzad Tokhmechi – School of Mining, Petroleum and Geophysics Engineering, Shahrood University of Technology
Ali Soleamani – School of Electrical and Robotic Engineering, Shahrood University of Technology, Shahrood, Iran

چکیده:

Image logs are resistivity or sonic tools with high resolution which are round in oil boreholes in order to detect rock properties. Fractures are one of the main rock properties which identified using image logs. As a usual, interpreters identify fractures from image logs subjectively. Extraction of fracture is indeed a complex problem: their contrast is often weak, their thickness is variable. In addition, a single natural fracture trace is composed in theborehole image of short and disconnected pieces and there are some others events that are similar to fracture such as bedding. In order to interpretfracture from image logs, a basic and complex step is to extract fractures from image logs. In this process, firstly, fractures and their similar eventssuch as bedding were extracted from image logs,and then, fractures were discriminated from similarevent with some features. In this paper, a novel technique is developed for fracture identification from image logs. In the proposed technique, after segmentation of image logs, some other image processing techniqueswere applied for extraction of fracture traces from other events. Morphological operations were applied on segmented image logs, in order to discriminating fracture pixels from other pixels. Segmented color image logs were converted to binary images. Usingopening morphological operation, intersected events were isolated from each others, and then, slop of components were used to remove horizontal andvertical traces which did not belong to fractures in image logs. In the final, erosion operation were applied repeatedly on these image to be extractedthe pixels of fractures from other pixel exactly. The final image were contains points which were belong to fractures. The proposed technique was applied for fracture detection on different image logs. Results revealed that application of competitive neural networks technique is better than others, and its results are compatible with conventional results, and using morphological operation can extract fractures from other their events exactly