基于岩屑定量数字化分析的吉木萨尔页岩油储层表征方法

页岩油由于储层异常致密，传统储层表征方法存在取心成本高、岩心测试周期长等问题，并且实验室测量方法对页岩储集空间、微观孔隙结构的精细刻画不足。为此提出了一种基于岩屑定量数字化分析的页岩油储层表征方法，通过对岩屑的扫描电子显微镜(SEM)图片进行基于深度学习的图像识别和分割，刻画出纳米级孔隙结构，再结合J函数分析方法，定量分析页岩油储层最低可动孔隙半径，用以辅助页岩油储层的高效开发。研究选取吉木萨尔芦草沟页岩油储层典型井吉174井岩石碎屑样品为研究对象，得到了该井页岩油储层可动用油的孔隙尺寸下限，同时核磁测井结果也验证了该方法的有效性。

Method characterizing shale oil reservoirs in Jimsar based on quantitative digital analysis of cuttings

For the reservoirs bearing shale oil, due to their abnormal tightness, there are many problems, such as high coring cost and long core testing period, in characterizing these reservoirs with traditional methods. Furthermore, it is also insufficient for laboratory measurement methods to finely characterize the shale reservoir space and microscopic pore structure. To this end, a method characterizing shale oil reservoirs based on quantitative digital analysis of cuttings is proposed. Based on deep learning, the scanning electron microscope (SEM) images of cuttings were recognized and segmented to characterize the nano-scale pore structure, and then combined with the J-function analysis method, the minimum movable pore radius of shale oil reservoirs was quantitatively analyzed to assist the efficient development of shale oil reservoirs. The study selects the cutting sample from Well Ji 174, a typical well in the Lucaogou shale oil reservoir in Jimsar, as the research object, and gets the lower limit of the pore size that can allow oil movement, in addition, the nuclear magnetic logging results also verified the effectiveness of the method.