Abstract
To address the absence of unified quantitative evaluation criteria for matching well pattern deployment with fracture networks in complex fractured reservoirs, and the reliance on empirical methods and high-cost numerical simulation traversal in well pattern optimization, a probabilistic statistics-based approach for statistical modeling and optimization of well pattern–fracture network matching is proposed. A statistical criterion for well pattern–fracture network matching is established, centered on "minimizing the variance of the expected value of the normalized effective fracture projection length in each injection–production direction." This criterion transforms the objective of achieving balanced displacement into a solvable variance minimization problem. On the basis of treating fracture length, orientation, and position as random variables, with their probabilistic characteristics described respectively by the exponential distribution, von Mises distribution, and uniform distribution, analytical expressions for the expected value and variance of the effective fracture projection length are obtained through rigorous mathematical derivation. These expressions enable quantitative evaluation of the degree of well pattern–fracture network matching, thereby converting an optimization problem conventionally addressed through numerical simulation traversal into one solvable via analytical functions. The proposed method is validated through optimization and numerical simulation of a five-spot well pattern. Results demonstrate that for the optimized well pattern, the expected values of effective fracture projection length in all directions are highly balanced, and the range of cumulative oil production decreases from 86,087 m3 to 11,642 m3, indicating a marked improvement in displacement uniformity and validating the effectiveness of the proposed statistical criterion and model.