An effective integration optimization algorithm for regional fracturing design and drilling placement

Abstract

Hydraulic fracturing horizontal well technology has a pivotal role in exploiting unconventional resources, but it often requires considerable investment. Integrated optimization of the fracturing design, horizontal well, and drilling platform placement in the whole region is an efficient way to reduce cost and increase economic benefits. However, it is difficult to quickly complete these optimizations because traditional reservoir numerical simulation requires extensive resources in terms of labor, time, and computation. Therefore, we proposed a new algorithm, named Gaussian Process Regression-Bilevel Programming Genetic-Critical Circle Intersection Momentum Shift (GPR-BPG-CCIMS) algorithm to effectively co-optimize the fracturing design, horizontal well and drilling platform placement in the whole region. BPG algorithm shares the merits of Bilevel Programming (BP) and genetic algorithm (GA), which is used for optimizing the horizontal well and fracturing parameters. CCIMS algorithm is applied to optimize the drilling platform placement by merging and shifting drilling platform positions. GPR model is a production proxy model based on machine learning to improve optimization efficiency. The proposed algorithm was applied to a real shale gas reservoir to verify the effectiveness of the algorithm. The results showed that the GPR model exhibited great prediction accuracy by providing the high coefficient of determination (R2 = 0.999), the high percentage of accuracy-precision (PAP = 99.017) and the low root mean square error (RMSE = 0.022). The optimal fracture half length, cluster spacing, fracture fluid per meter, horizontal section length, well location, and drilling platform location were obtained by GPR-BPG-CCIMS algorithm. Compared with the actual net present value (NPV), the optimized NPV was increased by 181 million dollars. The proposed algorithm successfully co-optimizes the engineering parameters of the whole region in a short consumption time, which effectively supports the development scheme design and decision-making. The optimization framework can be extended to other shale gas basins.