A History Matching Framework to Characterize Fracture Network and Reservoir Properties in Tight Oil
A History Matching Framework to Characterize Fracture Network and Reservoir Properties in Tight Oil
复制标题
表征致密油裂缝网络和储层性质的历史匹配框架
DOI：
10.1115/1.4044767
复制
发表时间：
2020
影响因子：
3
通讯作者：
Feng Qihong
中科院分区：
文献类型：
--
作者：
Xu Shiqian;Li Yuyao;Zhao Yu;Wang Sen;Feng Qihong
准确表征致密油藏水力裂缝网络和储层物性，可以为产量预测和开发设计奠定基础。在这项工作中，我们提出了一个致密油的历史匹配框架。首先利用Hough变换方法从微地震资料中刻画复杂裂隙网络。然后，将裂缝网络引入到嵌入式离散裂缝模型（EDFM）中，建立致密油藏数值模拟模型。在此基础上，进一步将鲸鱼优化算法（WOA）与EDFM相结合，对现场生产数据进行匹配。通过这种方式，我们可以准确地估计储层性质，包括基质渗透率和孔隙度，以及裂缝渗透率。我们将该框架应用于中国的两个领域的应用。一是大庆油田松辽盆地压裂直井。另一种是新疆吉木萨尔凹陷的多级压裂水平井。结果表明，如果我们不考虑致密油的特性，估计的裂缝渗透率，基质渗透率和基质孔隙度将分别低估73%，20%和47%。由于我们首次将WOA应用于历史匹配，因此我们将WOA的性能与具有多重数据同化的集合平滑器（ES-MDA）进行了比较。当我们拟合六个参数时，ES-MDA的性能优于WOA。然而，当我们拟合三个参数时，WOA比ES-MDA表现得更好。此外，对于工程问题，WOA在收敛速度和稳定性方面都有很好的表现。因此，在今后的历史拟合应用中，推荐使用WOA。
Accurately characterizing hydraulic fracture network and tight oil reservoir properties can lay the foundation for the production forecast and development design. In this work, we proposed a history matching framework for tight oil. We first use the Hough transform method to characterize complex fracture network from microseismic data. Then, we put the fracture network into an embedded discrete fracture model (EDFM) to build a tight oil reservoir simulation model. After that, we further couple whale optimization algorithm (WOA) and EDFM to match the field production data. In this way, we can accurately estimate reservoir properties, including matrix permeability and porosity, as well as fracture permeability. We apply the framework to two-field applications in China. One is fractured vertical well in the Songliao Basin of Daqing oilfield. The other one is multi-stage fractured horizontal well in the Jimsar Sag of the Xinjiang oilfield. Results show that if we do not consider tight oil characteristics, the estimated fracture permeability, matrix permeability, and matrix porosity will underestimate 73%, 20%, and 47%, respectively. Because we apply WOA to history matching for the first time, we compare the performance of WOA with ensemble–smoother with multiple data–assimilation (ES-MDA). When we fit six parameters, ES-MDA performs better than WOA. However, when we fit three parameters, WOA performs better than ES-MDA. In addition, for engineering problem, WOA performs well on both convergence speed and stability. Therefore, WOA is recommended in the future application of history matching.