天然气工业2025，Vol.45Issue(11)：111-121,11.DOI:10.3787/j.issn.1000-0976.2025.11.009
基于XGBoost和多目标优化的碳酸盐岩气藏水侵智能排采调控
Control of intelligent drainage-production in water-invaded carbonate gas reservoirs based on XGBoost and multi-objective optimization
摘要
Abstract
Uneven water invasion is a significant challenge that restricts the ultimate recovery of water-bearing carbonate gas reservoirs.Active drainage can effectively mitigate uneven water invasion damage,thus significantly enhancinge ultimate gas recovery.The drainage-production well pattern and system have direct impacts on the performance of water control.Optimizing drainage-production parameters to achieve the best development scheme is crucial for facilitating practical field applications.This paper takes the Cambrian Longwangmiao Formation carbonate gas reservoir in the Sichuan Basin as an example to discuss the optimization of drainage-production system in water-bearing carbonate gas reservoirs.A comprehensive evaluation model of drainage well was developed based on machine learning and was then used to determine the optimal drainage-production well pattern.Additionally,a multi-objective optimization method was utilized to obtain the optimal drainage-production system incorporating intelligent control.The following results are obtained.First,the main factors influencing the productions gas and water after uneven water invasion are water-gas ratio,permeability,well spacing,number of gas production wells,and porosity,in a descending order of influence.Second,considering these factors,a comprehensive evaluation model of drainage well is built.With this model,three water-flooded production wells are identified suitable for convention to drainage wells,and accordingly the optimal drainage-production well pattern is determined.Third,the multi-objective optimization method is used to obtain a global optimal drainage-production system,which facilitates the coordinated realization of two targets for maximizing cumulative gas production and minimizing cumulative water production from gas production wells.It is concluded that the optimization results align with the theoretical expectations,confirming the feasibility of the optimization method to quickly define the drainage-production system for water-bearing gas reservoirs.Moreover,the proposed method integrating machine learning,numerical simulation,and intelligent optimization algorithms enables intelligent control of uneven water invasion in carbonate gas reservoirs,which provides new insight for mitigating uneven water invasion in gas reservoirs and supports the leapfrog growth of natural gas production strongly.关键词
四川盆地/碳酸盐岩气藏/水侵调控/机器学习/排水井/多目标优化/智能排采Key words
Sichuan Basin/Carbonate gas reservoir/Water invasion control/Machine learning/Drainage well/Multi-objective optimization/Intelligent drainage-production分类
能源科技引用本文复制引用
赵玉龙,谢泽豪,张连进,张芮菡,李韬,张飞,骆锐柯,张烈辉..基于XGBoost和多目标优化的碳酸盐岩气藏水侵智能排采调控[J].天然气工业,2025,45(11):111-121,11.基金项目
国家自然科学基金联合基金重点项目"四川盆地有水气藏CO2驱提高采收率及其有效封存研究"(编号:U23A2022)、中国石油-西南石油大学创新联合体项目"多重介质跨尺度升级的有水气藏数值模拟技术"(编号:2020CX010403). （编号:U23A2022）
