Abstract: Deep coalbed methane (CBM)reservoirs have become an important growth driver for China’s natural gas production. Field practices have revealed that fracturing sand backflow frequently occurs during the wiper trip or workover operations in deep CBM wells. This phenomenon leads to issues such as tubing erosion, sand accumulation and blockage, and production decline, which directly affect the estimated ultimate recovery (EUR)of a single well. The variation in the conductivity of supporting fractures is therefore a critical factor influencing gas production performance in deep CBM reservoirs and plays a key role in establishing appropriate fluid drainage strategies. However, existing commercial simulators fail to incorporate the influence of velocity-sensitivity effects and thus cannot characterize proppant transport and fracture conductivity evolution during the drainage and production process. As a result, the production forecast accuracy for CBM wells remains low, making it difficult to establish a scientifically based and quantitatively optimized flowback strategy. To address these challenges, core flooding experiments and particle mechanics analyses were first conducted to investigate the influence mechanisms of factors such as sand concentration, proppant particle size, and effective stress on the velocity sensitivity characteristics of fracture conductivity in deep coal seams. Based on the experimental findings, a dynamic conductivity characterization model considering the effect of flow velocity was then developed. By coupling this model with a discrete fracture model and a dual-porosity single-permeability coal seam model, the study establishes a numerical simulation method considering the effects of fracture conductivity changes. This method ultimately enables the determination of a reasonable flowback strategy. The results indicate that:(1)With the increasing of flow velocity, the fracture conductivity in deep coal seams first rises and then declines, exhibiting a peak at the flow velocities between 2.5 and 7.5 mL/min. The variation amplitude of conductivity is positively correlated with proppant placement concentration, particle size, and fluid viscosity, while being negatively correlated with effective stress. (2)The intrinsic mechanism governing the dynamic variation of fracture conductivity in coal rock systems lies in the dual effect of fluid flow on proppant behavior. At low flow velocities, the flushing and sand-carrying action of the fluid on small-sized proppant particles exerts a positive effect by clearing and widening the flow channels. In contrast, at high flow velocities, the fluid tends to displace large-sized proppant particles that provide structural support, producing a negative effect by promoting fracture closure. (3)Numerical simulation results that incorporate the velocity sensitivity characteristics of fracture conductivity indicate that the velocity sensitivity effect significantly reduces gas production performance. Therefore, it is necessary to control the flowback rate to mitigate the adverse impact of velocity sensitivity effects. (4)The production practices in Daji block have verified that the fracturing sand backflow volume and gas production performance of CBM wells are both governed by the liquid production rate. Based on comprehensive analyses combining physical experiments, numerical simulations, and field monitoring data, it is recommended that the flowback rate for a single layer of vertical well should be maintained at 20-23 m³/d, with a maximum daily liquid production not exceeding 23 m³/d. For horizontal wells with the single-stage fracturing parameters and stimulation effects similar to those of vertical wells, the maximum daily liquid production should not exceed (number of stages × 23)m³/d. The field application in Daji Block demonstrates that the maximum daily liquid production of CBM horizontal wells has been effectively reduced from 634 m³/d to below 230 m³/d. As a result, the occurrence of fracturing sand backflow is significantly mitigated, and gas production performance is markedly improved. These findings provide both theoretical support and technical guidance for formulating a reasonable flowback strategy for deep CBM wells.

Key words: deep coalbed methane (CBM), coal-rock gas, velocity-sensitive characteristics, fracture conductivity, fracturing sand backflow, numerical simulation, production and drainage control, practical significance, Daji block

摘要： 深部煤层气已成为中国天然气增储上产的重要增长极。现场实践发现,深部煤层气井在通井或修井时普遍存在压裂砂返吐现象,造成管柱冲蚀磨损、积砂堵塞、产量下降等,直接影响单井预计最终可采储量,而支撑裂缝导流能力变化是影响深部煤层气产气效果和制定合理排液制度的关键。但现有商业模拟器未考虑速敏效应影响,无法刻画排采过程中支撑剂运移导致的导流能力动态变化特征,导致深部煤层气井产量预测精度差,难以科学制定定量化返排制度。为此,首先通过岩心实验和颗粒作用分析,探究了铺砂浓度、支撑剂粒径、有效应力等因素对深部煤层支撑裂缝导流能力速敏特征的影响机制,然后基于实验规律建立了考虑流速影响的动态导流能力表征模型,耦合该模型,嵌入离散裂缝模型与煤层双孔单渗模型,构建了考虑压裂缝导流能力变化影响的数值模拟方法,进而确定合理返排制度。研究结果表明:①随着流速增加,深部煤层支撑裂缝导流能力呈现先上升、后下降的变化趋势,导流能力峰值对应流速为2.5~7.5 mL/min,且导流能力变化幅度与铺砂浓度、粒径和流体黏度正相关,与有效应力负相关。②煤岩支撑裂缝导流能力动态变化的内控机制在于,低流速下液流对小粒径支撑剂的冲刷携砂作用具有疏通流动通道的正向效应,而高流速下液流将起支撑作用的部分大粒径支撑剂冲刷出,具有闭合裂缝的负向效应。③考虑支撑裂缝导流能力速敏效应差异的数值模拟结果表明,速敏效应会显著影响产气效果,需通过控制返排液量以降低速敏效应的不利影响。④大吉区块生产实践证实了深部煤层气井支撑剂返吐量和产气效果受产液量影响显著,基于物理实验、数值模拟及现场监测数据统计分析,多种方法综合确定直井单层返排液量为20~23 m³/d,最大产液量不超过23 m³/d,水平井单段压裂参数和改造效果与直井相似时,建议水平井最大产液量不超过(段数×23)m³/d。研 究成果应用于大吉区块深部煤层气水平井返排液量控制,最高产液量由634 m³/d普遍控制至230 m³/d以下,压裂砂返吐现象明显减少,产气效果显著提升,为深部煤层气井合理返排制度提供了理论依据和技术支撑。

关键词: 深部煤层气, 煤岩气, 速敏特征, 导流能力, 压裂砂返吐, 数值模拟, 排采控制, 实践意义, 大吉区块