Fracture Classification-Grading Prediction Technology and Application in Carbonate Reservoir Rocks:A Case Study from Tahe Oilfield,Tarim Basin
Wang Zhen
Rao Ying
Xie Haiyang
Shi Guanzhong
Abstract：Carbonate reservoirs are important carriers of global oil and gas resources.The development characteristics of their internal karst caves and fault systems directly affect the storage and migration capacity of oil and gas.In order to solve the problem of multi-scale fault prediction in deep and ultra-deep carbonate reservoirs,this study takes the Ordovician carbonate rocks in the Tahe oilfield in the Tarim basin as an example and proposes a classification and grading fault prediction technology based on seismic wave field characteristic analysis.The interference mechanism of large-scale fracture-cavity bodies on conventional fault prediction attributes(such as coherence and maximum likelihood)is revealed through three-dimensional forward simulation,and it is found that the"beaded"reflection anomaly at the boundary of the fracture-cavity body will lead to false connectivity of faults and relocation deviation.Based on the differences in the scale and dissolution characteristics of the faults,the fault system in the study area is divided into large-scale broken-dissolution faults(>20 m),medium-scale weak-undissolved faults(10-20 m)and small-scale fractures(<10 m),and targeted prediction methods are developed for each of them:for large-scale faults,a fracture retrieval technology based on gradient structural tensor thinning is proposed to effectively overcome the interference of abnormal boundaries of karst caves;for medium-scale faults,the longitudinal continuity of the faults is significantly improved by combining AFE coherent enhancement attributes with U-Net deep learning algorithm;for small-scale fractures,the Likelihood attribute and structural guidance filtering are used to accurately extract weak reflection signals.Further,a well control multi-attribute fusion model is constructed by fusing multi-scale fault attributes and drilling loss data through deep feed forward neural network(DFNN).The application results show that this technology system has achieved full-scale characterization of the fault system in the complex fracture-cave area of Tahe oilfield.Large-scale strike-slip faults are distributed in a conjugate NNE-NNW direction,medium-scale faults form flower-like structures,and small-scale fractures are densely developed on the active disk(east side)of the fault.This study provides a new technical approach for the prediction of deep carbonate reservoir faults and has important reference value for the exploration and development of similar oil and gas reservoirs.
Keywords：carbonate rockfracture classificationstructural tensorseismic forward modelingTahe oilfieldpetroleum geology
Publication Date：2025-12-30
Online Publishing Date：2026-01-26(First online date of this platform, not the publication date of the document)
Pages：19( 4764-4782 )