鄂尔多斯盆地中、东部深层煤岩气水平井高效开发主控因素

Main factors controlling the efficient production of horizontal wells for deep coal-rock gas in the eastern and central Ordos Basin

费世祥 崔越华 李小锋 汪淑洁 王晔 张正涛 孟培龙 郑小鹏 徐运动 高建文 罗文琴 蒋婷婷

Shixiang FEI Yuehua CUI Xiaofeng LI Shujie WANG Ye WANG Zhengtao ZHANG Peilong MENG Xiaopeng ZHENG Yundong XU Jianwen GAO Wenqin LUO Tingting JIANG

鄂尔多斯盆地深层煤岩气资源丰富，具有巨大的开发潜力和良好的发展前景。但深层煤岩气地质特征区域变化较快，勘探评价阶段开展了不同工艺条件下的改造试验，气井产能呈现出较大的差异特征。为了揭示影响煤岩气水平井产能的主控因素，充分利用先导试验区深层煤岩气动态和静态资料，按照地质-工程一体化融合分析思路，精细描述煤岩地质特征，并精细评价投产井的生产指标。采用皮尔逊相关性分析、系统聚类和机器学习等方法，对不同地质工程因素进行量化评价。研究结果表明：①在煤岩结构和热演化成熟度等区域地质特征相近条件下，地质因素中煤岩厚度和含气性对煤岩气产能影响显著，工程因素中钻遇煤岩长度（）、总液量（）、加砂量（）和加砂强度（）与产能正相关性较好。②工程因素与首年日产气量相关性明显优于地质因素，且地质-工程复合因子相关性明显优于单因素，增加井控体积和改造规模有利于气井高产。③在水平井开发中，当目标层煤岩厚度为6 ~ 10 m、平均加砂强度为5.5 t/m时，为实现水平井单井最终累计产气量（）达到5 000 × 10 m，需要的水平段长度为1 000 ~ 1 500 m。④综合利用深度神经网络、支持向量机及随机森林模型等创建了基于地质及工艺参数的深层煤岩气单井产能预测新方法，采用22口井盲井验证，预测符合率高达91 %。该研究成果对深层煤岩气目标区优选及开发方案的编制具有重要的指导作用，对中国其他煤岩气区块开发技术对策优化也具有借鉴意义。

Abundant deep coal-rock gas in the Ordos Basin boasts enormous potential for exploration and development. The geological characteristics of deep coal-rock gas exhibit substantial regional changes， and reservoir simulation tests under various process conditions have revealed notable differences in the productivity of coal-rock gas wells. This study aims to investigate the main factors controlling the coal-rock gas productivity of horizontal wells. Using the dynamic and static data from deep coal-rock gas in the pilot test area and employing the analytical approach of geology-engineering integration， we depict the geological features of coals in detail and thoroughly assess the production indices of producing wells. Furthermore， geological and engineering assessments are carried out using methods such as Pearson correlation analysis， hierarchical clustering， and machine learning. The results indicate that under similar regional geological features such as coal structure and thermal maturity， coal thickness and gas-bearing property， two geological factors， significantly affect the productivity of coal-rock gas wells. Meanwhile， the total length of coals encountered during drilling （）， total drill-in liquid volume （）， proppant volume （）， and proppant intensity （） among engineering factors exhibit positive correlations with the productivity of coal-rock gas wells. Notably， engineering factors show more pronounced correlations with the first-year daily gas production compared to geological factors， and the composite geology-engineering factors display more significant correlations than individual factors. Increasing well-controlled drainage volume and stimulated reservoir volume （SRV） contributes to the high productivity of coal-rock gas wells. Assuming target coal thicknesses measuring 6 ~ 10 m and an average proppant intensity of 5.5 t/m， a lateral length of 1 000 ~ 1 500 m is required to achieve a single-well estimated ultimate recovery （EUR） of 5 000 × 10 m. A new method for predicting the single-well productivity of deep coal-rock gas， based on geological and engineering parameters， is developed using intelligent algorithms such as deep neural networks， support vector machines （SVMs）， and random forest models. This method achieves a coincidence rate of up to 91 % according to blind well verification with 22 wells involved.

10.11743/ogg20250119