Study On The Methods Of Identification And Description For Carbonate Fractures

The development of carbonate fractures is often rather complex and high heterogeneity,so there are some obstacles in fracture description. It exists some problems, such as the macroscopical fractures investigation of field and core,the well logging appraisal of validity, the fracture integrated identification of conventional logging technique, the period research of fractures by means of microscopical laboratory facilities and so on. With regards to these problems, the paper investigates the correlation description methods of carbonate fractures.Based on the actual survey of field and core fractures, the macroscopic features and control factors of. carbonate fractures are analyzed intuitively from plane and section perspective in the form of many typical photos and sketch maps, and the description methods of field and core fractures are evaluated. These help to guide the fracture recognition in surface and subsurface. By the combinations of a considerable amount of core and imaging logging data, the paper analyzes the response characteristic of carbonate fractures in imaging logging. According to the feature of all kinds of logging methods, it also researches the evaluation methodology of validity by the comprehensive utilization of new well logging technologies and introduces the quantitative description of fracture utilizing FMI logging data. In order to make full use of the conventional logging materials to discern the fractures widely, the paper employs the neural network theory, designs three-layer BP network and programs recognition procedures by Matlab language. The procedure adopts improved BP algorithm and LM algorithm to deal with characteristic parameter data extracting from conventional logs and computes the unknown samples through the learning of known samples and the model building. A comparison of the recognition results and imaging logging data reaches conclusion that the neural network can be used in the fracture recognition and obtain perfect results. A comparison of improved BP algorithm and LM algorithm discovers that the latter algorithm is better than the former. LM algorithm is practical and effective back propagation training algorithm. With reference to the period research of fractures, the paper analyzes the features of fracture filling in terms of the stable isotopic, fluid inclusion, cathode light and florescent light according to many oilfield examples, and investigate the classification methods of fractures.