Linear flow simulated by traditional productivity simulation method when determining gas reservoir property cutoffs is inconsistent with radial flow in actual field single well production, and the assumption that gas velocity at outlet of core is equal to gas velocity in field wellbore is inconsistent with similarity criterion required by physical model. By improving experiment method for linear flow, i.e. radial flow simulation method, new experimental equipment is designed to simulate areal radial flow which is more in line with actual streamline of gas reservoir. Gas flow rate obtained from lab radial flow experiment is transformed to field conditions according to similarity criterion so as to more realistically evaluate flow capacity of target reservoir and finally obtain property cutoffs that conform to productivity characteristics of target gas reservoir. The results show that gas well productivity obtained by improved radial flow productivity simulation method is increased by 100 times higher than that of linear flow productivity simulation method before improvement. Property cutoffs obtained by improved radial flow productivity simulation method is lower, which is more in line with actual production situation with low porosity and low permeability but high single well productivity of ultra-deep carbonate gas reservoir in Member 4 of Dengying Formation in Gaoshiti-Moxi Block. The research results can more accurately evaluate property cutoffs of ultra-deep carbonate gas reservoirs.
In order to overcome the evaluation distortion of conventional evaluation method for retrograde condensate damage, which ignores near-wellbore retrograde condensate superposition effect and far-wellbore single phase condensate gas flow characteristics, based on long-core displacement experiment, a new evaluation method is established considering superposition effect and far-well single-phase flow, simulating retrograde condensate damage of shale reservoir in actual depletion development and evaluating retrograde condensate removing effect of the injections of methanol, dry gas and CO 2 . Influence of CO 2 injection volume and injection pressure on evaluation indexes of retrograde condensate damage is studied. Experiment results show that, when the pressure drops to the maximum retrograde condensate pressure, gas permeability damage rate is 80.2%～83.1%, with displacement pressure difference rapidly increasing from 5 MPa to 15 MPa and gas flow resistance increasing by 200%. After CO 2 injection, gas permeability increases by 3.2 times with gas-oil ratio changing rate of 23.1% and core differential pressure decreasing rate of 36.1%, verifying that CO 2 injection has the best effect, followed by methanol, and dry gas injection has the worst effect. The optimal injection pressure of CO 2 is 9 MPa with injection volume of 0.15 PV and oil-gas equilibrium time of 9.9 h. The research provides ideas and methods for efficient development of shale condensate gas reservoirs and retrograde condensate damage removal.
目前已有关于裂缝的研究主要是地震单属性预测,裂缝预测的精度较低.为了进一步提高压裂改造水平与页岩气井产能,需对页岩储层不同尺度的裂缝进行预测,尤其是对小尺度裂缝进行精细预测.利用相干、曲率、蚂蚁体3种地震属性对永川地区龙马溪组一段的天然裂缝进行定量预测,并针对不同尺度裂缝进行多属性裂缝综合建模.结果表明:永川地区裂缝分为3个等级,断距大于100 m的大尺度裂缝、断距50～100 m的中尺度裂缝和断距小于50 m的小尺度裂缝,相干、曲率、蚂蚁体属性预测结果大体一致,整体上断层走向以NE—SW为主,大尺度裂缝主要发育在新店子背斜附近,蚂蚁体可以更清晰地刻画出小尺度裂缝.裂缝综合建模显示,永川地区深层页岩气储层总体裂缝发育程度由好到差依次为新店子背斜主体、南部向斜、北区向斜,南部向斜微裂缝最发育,有利于压裂改造,因此将永川南部作为首要滚动产建目标区,气井试采效果好,测试产量整体高于永川中、北区.研究成果对于实现永川地区页岩气增储上产具有一定的指导意义.
为了解决页岩油组合粒径+滑溜水的支撑剂加砂工艺中裂缝有效支撑差、导流能力弱的问题,建立支撑剂粒径分布的稠密离散相模型(DDPM),研究压裂主缝中组合粒径支撑剂加砂运移及铺置规律,并基于运移规律模拟结果,开展劈裂页岩岩板组合粒径不同铺置模式下的室内导流能力评价.结果表明:滑溜水携砂液体系下,裂缝内支撑剂叠置铺置时,后注入的支撑剂叠置于先注入支撑剂的顶端,且先注入的支撑剂会被后续注入的支撑剂向远端推移一定距离;组合粒径中粒径配比差异对于支撑剂运移形成的砂堤形态影响较小;大粒径组合逐级注入的方式更利于支撑剂在近缝口和裂缝内垂向铺置;在低闭合压力(p≤40 MPa)、铺砂浓度5kg/m2条件下,沉降铺置方式最利于提高裂缝导流能力,其次为混合铺置,分段铺置方式最差;高闭合压力下(p＞40MPa),铺置方式对裂缝导流能力影响较弱.综合支撑剂运移模拟和导流能力评价结果,建议吉木萨尔页岩油组合粒径加砂工艺采用逐级注入的方式,并保证组合粒径中大粒径拥有较大配比.
砂地比是表征储层侧向输导能力的重要参数,基于线性回归方法结合地震属性预测砂地比易受数据间非线性关系影响,且样本量较少,预测误差较大.以束鹿凹陷束21井区为例,研究了馆陶组砂地比平面分布趋势,通过细分层方法扩充数据样本规模及平滑因子滤波手段对地震属性进行"毛刺"平滑处理,采用皮尔逊相关性优选地震属性,结合多种地震属性和井点砂地比统计数据,使用随机森林算法预测砂地比.结果表明,随机森林法可以有效处理非线性结构数据,与线性回归、神经网络(MLP)对比,模型误差明显降低,预测精度可达80％.该方法的预测结果已成功应用于实际钻井中,且取得了显著的效果.
