Comprehensive Evaluation of Fractures in Ultra‐Deep Basement Gas Reservoirs: A Case Study From the Northern Qaidam Basin

Fractures are key factors influencing the formation of ultra‐deep basement gas reservoirs and the productivity of gas wells. Therefore, establishing a comprehensive fracture evaluation method suitable for ultra‐deep basement gas reservoirs is of great significance for guiding their exploration and development. This study takes the Devonian basement reservoir in the K2 well block of the northern Qaidam Basin as an example, systematically investigates the fracture characteristics of ultra‐deep basement gas reservoirs by integrating core data, FMI (Formation MicroImager) logging and seismic data, and proposes a multi‐scale fracture modelling approach. The results show that the fracture linear density in the basement reservoirs of the study area typically ranges from 0.5 to 6 strips/m, with a certain negative correlation between single‐layer thickness and fracture density. When the thickness of a single lithological layer exceeds 10 m, the fracture density is generally less than 1 fracture/m. Based on sensitivity analysis of basement fractures, a comprehensive fracture index ( F D ) was constructed for fracture identification in basement reservoirs. Under the constraints of single‐well fracture development and orientation statistics, large‐scale fractures in the basement formations of the study area were extracted using seismic fracture anti‐tracking attributes. A discrete fracture network (DFN) modelling method with multi‐parameter constraints was employed to systematically integrate geological information such as anti‐tracking attributes, distance to faults and structural curvature. Then, combined with fracture statistics from single‐well FMI logging, a multi‐constrained modelling approach was used to establish a small‐to‐large‐scale DFN model consistent with geological laws. Finally, a multi‐scale fusion DFN modelling technique was applied to achieve 3D coupling of large‐ and small‐scale fracture systems within a unified grid framework. The consistency rate between well‐log‐interpreted and simulated fracture linear densities in the basement formations exceeds 93%, demonstrating the reliability of the fracture prediction results.