A study on grille structure modeling algorithm for fault-controlled fractured-cavity reservoirs: A case study of the shunbei no. 5 fault zone

Abstract

Establishing high-resolution 3D geological models of fault-controlled reservoirs is crucial for optimizing well placement and development plan. Carbonate fault-controlled fracture-cavity reservoirs in the Shunbei area of the Tarim Basin, Northwest China, exhibit complex heterogeneity. These reservoirs typically comprise fracture planes, caves and disordered bodies. Fracture planes are narrow and banded, with caves and disordered bodies distributed around them. Within fracture planes and caves, crush belts and bedrock belts alternate to form grille structures. These pose significant challenges for traditional modeling algorithms to characterize accurately. To address this, we proposed a hierarchical object-based modeling algorithm to reproduce fault-controlled fracture-cavity body's grille structural trends and shapes. Using seismic data from the Shunbei No.5 Fault Zone (with a resolution of 25∗25m) and well logging data (including drilling fluid loss data and resistivity logging data), conduct research on grille structure modeling algorithms. First, the fault-controlled fracture-cavity reservoirs are distinguished by fracture planes, caves, and disordered bodies, and contour models are established via seismic attributes threshold truncation. Second, statistics on the scale of development of crush belts and breccia belts under 100 m of fracture planes and caves in different stress sections by logging data. A regional growth tracking algorithm are applied to identify fracture planes trend lines, which can be classified into single, multi, convergent, and branching forms based on contour characteristics. Third, cumulative probability sampling is used to determine the number and scale of the crush and breccia belts. Grille structure models were constructed at three levels: bedrock, crush, and breccia belts. Results indicate successful identification of trend lines matching the structural contours, establishing accurate grille structure models by employing hierarchical simulation strategy under trend line constraints. The models established by traditional methods exhibit significant randomness, making it difficult to control both the variable developmental trajectories of individual belts and the relative positional relationships among multiple belts. Based on these geological facies models, corresponding physical property models were generated, achieving high accuracy in reserve calculations and numerical simulations with less than 10 % error, thus providing valuable guidance for oil and gas development. In the future, more compatible contour models can be established through methods like multi-attribute fusion and deep learning. By integrating production data, the developmental positions and connectivity of grille belts can be constrained.