鄂尔多斯盆地东缘神府深煤层大气田探明地质储量超千亿方，实现该地区深部煤层气高效开发对保障国家能源供应具有重要意义。但是，由于深煤层地质环境复杂，具有高地应力、中高温度、特低渗透、强非均质性、割理/裂隙发育等特点，导致现有中浅煤层压裂改造技术难以完全适用于深煤层，其施工规模与参数仍处于探索阶段。为了探究与深煤层地质条件相适应的增产改造技术，以鄂尔多斯盆地神府区块为地质背景，以深煤层大规模体积压裂为工程实践，围绕“极限动用+均衡扩展+有效支撑”的设计理念，提出“少段多簇适度密切割+等孔径深穿透限流射孔+复合液造缝(高黏液体破岩+低黏液体造复杂缝)+大排量高强度加砂+前置酸液降低破裂压力+多粒径组合支撑剂”为核心的体积压裂技术，并引入地质−工程−智能一体化压后评估方法，通过压裂−产能双重智能拟合校正，精细刻画了储层改造体积(SRV)和气体泄流体积(DRV)，预测了不同压裂规模及井型条件下的最终可采储量EUR。最后，通过统计神府区块深煤层压裂井地质、工程和产量特征数据，采用随机森林算法量化分析了影响深部煤层气产能的主控因素。实践结果表明：采用上述大规模体积压裂技术，已投产的直/定向井最高日产气量超过1×10⁴ m³，水平井最高日产气量超过2×10⁴ m³，说明深煤层可压性良好、开发潜力巨大；深部煤层气峰值产气量的主要影响因素为：煤层含气量、煤层厚度和加砂强度；累积产气量的主要影响因素为：煤层含气量、加砂强度和总砂量。

The proven geological reserves of the Shenfu deep coalbed methane (CBM) field on the eastern margin of the Ordos Basin exceed 100 billion cubic meters. It is of great significance to realizing the efficient development of deep CBM in this region to ensure the national energy supply. However, the complexity of the geological environment which includes high stress, medium-high temperatures, low permeability, strong heterogeneity, and wide developed cleats and natural fractures, makes it challenging for the existed shallow and medium CBM fracturing techniques to be fully applicable to deep CBM resources. As a result, the stimulation scale and parameters for deep coalbed fracturing are still in the trial-and-error stage. In order to explore the stimulation techniques which are compatible with the geological conditions of deep coalbeds, the Shenfu block in the Ordos Basin was taken as the geological background and the large-scale hydraulic fracturing of deep coal seams was conducted as an engineering practice. The authors designed the idea of “Push the limit to the beyond + balanced propagation + effective support”, and proposed the massive hydraulic fracturing techniques based on “multi-stage multi-clusters with moderate-dense cutting + perforation with equal apertures, deep penetration and limited flow + integrated variable viscosity (rock breaking by higher viscous slick water + complex fracture network generating by lower viscous slick water) + high pumping rate with high proppant concentration + pre-acid treatment to reduce the breakdown pressure + graded proppants with multiple sizes to support fractures”. Then, the authors put forward an integrated “Geology-Engineering-AI” workflow to perform post-frac analysis, through double matching and correcting the fracturing pumping pressure and production rate automatically, accurately characterized the stimulated reservoir volume (SRV) and drained rock volume (DRV), and predicted the estimated ultimate recovery (EUR) under different fracturing scales and well types. Finally, by statistically analyzing the gas production characteristics of multiple wells in the Shenfu block and utilizing the random forest method, the primary controlling factors affecting the production capacity of deep CBM were quantitatively analyzed. The results demonstrate that after reservoir stimulation, directional wells can achieve a maximum daily gas production rate exceeding 10 000 m³/d, while horizontal wells can achieve a maximum daily gas production rate exceeding 20 000 m³/d. It indicates that the deep coal beds have good fracturing properties and great development potential. The primary impact factors for peak gas production rate are gas content, coalbed thickness and proppant concentration, while the major impact factors for cumulative gas production include gas content, proppant concentration, and total volume of proppants.