Horizontal drilling in combination with hydraulic fracturing has been successfully used to efficiently and effectively exploit tight oil/ gas reservoirs where a multibranched fracture network may be generated near a horizontal well which has been confirmed with microseismic events (MSE). Due to their inherent constraints, limited attempts have been made to characterize such complex fracture networks at an individual fracturing stage using tracer flowback tests. In this work, numerical models have been developed, validated, and applied to describe the tracer flowback behavior and characterize the complex fractal-like discrete fracture networks in a tight formation. To be specific, the embedded discrete fracture model (EDFM) is employed to accurately capture the fracture geometry by first discretizing the fractures into various segments and then incorporating such discretized fractures into the matrix. The perpendicular bisector (PEBI) grids are then generated to flexibly conform to the fractures and reduce the grid orientation effects. Subsequently, a reservoir with complex fracture networks is discretized into two separate domains (i.e. matrix and fracture), while nonneighboring connections (NNCs) which can provide fluid communications between the two domains are applied to couple the matrix and embedded fractures. Furthermore, the tracer flowback profiles, which can reveal the complexity of fractal-like discrete fracture networks, are quantified by considering tracer advection, dispersion, and adsorption. Sensitivity analysis has been conducted to examine the effects of iteration number, branch number, deviation angle, scale factor, and the ratio of permeabilities of two cluster fractures on the tracer flowback concentration curves. It is found that the iteration number will greatly affect the tracer flowback concentration, while the deviation angle imposes a minor effect on tracer flowback behavior. An increase in both the branch number and scale factor will decrease the tracer flowback concentration. Two concentration peaks will appear when the permeabilities of the two cluster fractures are different. The larger the difference between the permeabilities of the two cluster fractures is, the greater the difference in the peak arrival time and the peak amplitude will be. In addition, the newly proposed model was verified and then applied to a field case to characterize the complex fractal-like fracture networks, which are confirmed with the microseismic events.

水平钻井与水力压裂相结合已成功地用于高效和有效地开采致密油气藏，其中可能在水平井附近产生多分支裂缝网络，这已通过微震事件 (MSE) 得到证实。由于其固有的限制，已经进行了有限的尝试来使用示踪剂回流测试在单个压裂阶段表征这种复杂的裂缝网络。在这项工作中，已经开发、验证并应用了数值模型来描述示踪剂回流行为和表征致密地层中复杂的分形状离散裂缝网络。再具体一点，嵌入式离散裂缝模型 (EDFM) 用于通过首先将裂缝离散为各个段，然后将这些离散裂缝合并到矩阵中来准确捕获裂缝几何形状。然后生成垂直平分线（PEBI）网格以灵活地顺应裂缝并减少网格定向效应。随后，具有复杂裂缝网络的储层被离散为两个独立的域（即基质和裂缝），而可以在两个域之间提供流体连通的非相邻连接（NNC）被应用于耦合基质和嵌入的裂缝。此外，通过考虑示踪剂平流、分散和吸附来量化示踪剂回流剖面，它可以揭示分形状离散裂缝网络的复杂性。进行了敏感性分析，考察了迭代次数、分支数、偏差角、比例因子以及两个簇状裂缝的渗透率比对示踪剂返排浓度曲线的影响。发现迭代次数对示踪剂返排浓度的影响较大，而偏离角对示踪剂返排行为的影响较小。支路数和比例因子的增加都会降低示踪剂回流浓度。当两个簇状裂缝的渗透率不同时，会出现两个浓度峰。两个簇状裂缝的渗透率差异越大，峰值到达时间和峰值幅度的差异就越大。此外，