Multi-Constraint Intelligent Optimization for Accelerated Horizontal Well Pattern Deployment in Complex Unconventional Reservoirs

Abstract The development of shale gas reservoirs in China's Sichuan Basin faces multiple constraints, including dense surface obstacles, complex fault systems, and significant interference from existing wells. Conventional deployment methods suffer from poor multi-constraint coupling, high dependence on operator experience, and low iteration efficiency, making large-scale, efficient development difficult. This study proposes an intelligent horizontal well pattern deployment algorithm (iPlan) tailored for complex formations. The proposed approach consists of a three-stage workflow. First, multi-source data from a three-dimensional geological model are integrated to accurately characterize the subsurface complexity, while various input parameters enhance system flexibility. Second, a rapid economic benefit evaluation model is established, employing sensitivity analysis to quantitatively assess the nonlinear effects of input parameters on recovery factor and net present value, thereby delineating the key parameter feasible domain. Finally, a fusion strategy combining genetic algorithms and spatial topology is adopted for the optimal well pattern deployment search. Based on the existing geological model of the Sichuan shale gas field, the system extracted structural, property, and geomechanical data while acquiring obstacle information including faults, existing wells, and surface villages. An economic model was employed to determine the optimal length of horizontal section and well spacing. Through adaptive penalty functions, the system transformed the spatial constraints into a value-constrained plane for well pattern deployment optimization. By integrating the geological-engineering sweet spot of the target layer, a constrained sweet spot map for well pattern deployment was generated to guide the well pattern deployment in autonomously avoiding high-risk zones and prioritizing the coverage of high-value targets. The system successfully optimized the deployment of 198 horizontal wells within 40 minutes. The optimized well pattern effectively avoided interference from 76 faults and 44 existing wells, achieving a 21% increase in the well control area and 17% improvement in the expected internal rate of return (IRR). The findings indicate that the method significantly reduces reliance on manual expertise and accelerates the iterative process, achieving an efficiency nearly 100 times higher than conventional approaches. These findings confirm the proposed intelligent algorithm successfully resolves the optimization challenges arising from complex constraints, enabling efficient horizontal well pattern deployment.