基于地质力学与机器学习的压裂诱发裂缝滑移及套管变形预测方法

Numerous casings deformed during hydraulic fracturing, which affected normal fracturing and production. Hydraulic fracturing could induce fracture slip, and the fracture slip will cause the shear action on casing that intersects with the sliding plane. It is significant to predict fracturing-induced slip and casing deformation in order to ensure shale-gas-well integrity. Firstly, the thermos-hydro-mechanical (THM) coupled model of fractured formation and well is established. It can simulate fracture slip and casing deformation. Secondly, the test of injection-induced slip is performed to validate the geomechanical model. Finally, the simulated results of casing deformation under various completion, fracturing and geology conditions compose a “big data” sample. By using machine learning (ML), the prediction model of casing integrity during hydraulic fracturing is built. Moreover, an intelligent system for casing integrity assessment is developed. It can alarm casing failure in time, and prevent casing failure by optimally designing the completion and fracturing parameters. The research findings can be applied to predict casing deformation quickly, which provide the theoretical basis for ensuring well integrity during fracturing.

水力压裂引起了大量的页岩气井套管变形，严重影响了正常压裂和单井产量。水力压裂诱发天然裂缝等弱面滑移，弱面滑移将对与之相交的套管产生剪切作用。定量预测压裂诱发裂缝滑移及套管变形对保障页岩气井井筒完整性具有重要意义。首先，建立裂缝性地层与井筒多场耦合模型，模拟水力压裂作用下裂缝滑移和套管变形等行为。然后，通过岩心注水剪切三轴试验，对地质力学模型进行验证。最后，由不同完井、压裂和地质条件下套管变形模拟结果组成“大数据”，利用机器学习算法建立压裂过程套管完整性预测模型，并开发完整性智能评价系统，预警页岩气井套管失效；对于评价为失效的井，优化设计完井和压裂参数，从而有效地预防压裂过程套管失效。研究成果可用于快速准确地预测压裂诱发套管变形，为保障压裂过程井筒完整性提供理论依据。

水平井压裂是经济有效地开发页岩气、致密油气和地热等非常规资源的重要技术手段，但会对井筒完整性造成威胁。页岩气井在压裂过程中套管变形(套变)问题频发，套变率高达约30%-50%。套变容易造成井下工具阻卡，增加了压裂作业的周期和成本，降低了压裂效果和单井产量。由于地质环境、压裂操作和裂缝扩展等复杂性，页岩气井套变问题是一个棘手的技术难题。.本项目提出了压裂套变智能预测方法，研究内容包括：建立压裂诱发裂缝滑移及套管变形的地质力学模型，分析水力压裂作用下裂缝与套管的力学行为；开展岩心注水剪切三轴试验，测量泵压、裂缝滑移和套管变形等压裂响应特征参数；研究基于机器学习的压裂过程套管完整性预测模型，开发完整性智能评价系统，实现早期失效预警和优化设计。.取得的研究成果：（1）通过页岩压裂套变模拟，揭示了套管剪切变形的机理，水力压裂诱发天然裂缝或断层滑移，地层运移对套管产生剪切载荷。（2）岩心注水剪切实验验证了假说和数值模拟，压裂液流入弱面，弱面滑移并剪切套管。（3）阐明了地质和工程因素对裂缝滑移及套变的影响规律：随着压裂液量、地应力差增加，裂缝滑移的位移增大；与水平最小地应力夹角30-60°的天然裂缝滑移较大；井筒远离天然裂缝、多簇控液量压裂可以降低裂缝滑移和套变的风险。（4）随机森林模型表明地质和工程因素对套变的影响程度相近，应综合考虑地质和工程因素进行套变预测与防控；模拟退火算法能够智能优化设计压裂施工参数以控制套变；采用Python语言开发了压裂套变智能预测软件。还取得了学术及人才成果：发表论文6篇；授权专利1件；软件著作权3件；培养研究生2名、本科生3名；获得省部级奖励2项；拟出版专著1部。.本项目构建了压裂过程页岩气井套管完整性评价方法，实现了套变智能预测，对套管完整性控制具有实用价值；深化了压裂过程地层及地下结构响应的认识；对智能钻完井发展具有推动作用。