This study focused on the model order reduction technique to improve the calculation speed of a traditional reservoir simulator.Considering that when the technique uses proper orthogonal decomposition (POD) method,it can accelerate the simulator but sensitive to system input parameters,thus leading to the reduction of the simulation precision and efficiency,a Krylov enhanced POD (KPOD) based on Krylov subspace and POD,was applied to the technique,with the aim of combining the moment matching property of Arnoldi with POD's data generalization ability to alleviate POD's dependence on input conditions.The case study of real reservoir simulation demonstrated that the KPOD outperforms POD in the calculation speed and accuracy,and proved the soundness of the method.%应用模型降阶技术进行了提高传统油藏模拟器运算速度的研究.考虑到该技术采用本征正交分解(POD)方法能够加速模拟器运算,但对系统的输入参数较敏感,降低模拟的精度和效率,采用了基于Krylov子空间和POD的KPOD方法,以便利用Amoldi的矩匹配性质和POD的数据泛化能力减少POD方法对模型输入参数的依赖.油藏模拟实例表明,KPOD方法在计算速度和精度上都优于POD方法,验证了该方法的有效性和实用性.
展开▼
