基于单井敏感性局域化EnKF的油藏辅助历史拟合方法

常用的油藏自动历史拟合算法多数存在梯度计算不准确、产生伪相关性等问题,导致参数修正错误、模型反演失真。通过建立一种基于单井动态敏感性的局域化集合卡尔曼滤波（FMM-CL-EnKF）历史拟合方法,解决了传统距离截断方法处理伪相关性时与实际地层状况不匹配的问题。基于程函方程,根据地质模型静态参数场信息,快速追踪压力波从井点到地层网格的传播时间确定各单井动态最大敏感性区域,从而构建局域化矩阵。同时,结合集合卡尔曼滤波（EnKF）方法,实现数据同化方法梯度的矫正,减弱伪相关,通过逐步拟合生产动态达到更新油藏模型和获取最优估计的目的。概念算例和矿场算例的计算结果表明,FMM-CL-EnKF方法在模型集合生产动态拟合效果及反演模型参数场准确性等方面均优于标准EnKF方法。     

EnKF整合三维地震数据和动态数据的应用

主要介绍集合卡尔曼滤波EnKF在更新油藏储层静态参数方面的应用。根据集合卡尔曼滤波产生的必要条件,介绍了集合卡尔曼滤波整合动态数据和三维地震数据的基本原理。通过实例分析,验证了集合卡尔曼滤波在整合动态数据和三维地震数据更新油藏参数方面的有效性,更新后的油藏模型能够较好地反映储层非均质性,并且与三维地震数据有较好的一致性。还比较了利用三维地震数据和四维地震数据进行更新油藏模型的差别。通过bKF方法将地震数据和动态数据结合起来描述油藏特征,能够很好地拟合观测数据,并获得较好的模型。     

协方差局地化方法在自动历史拟合中的应用

集合卡尔曼滤波(EnKF)是自动历史拟合领域应用较为广泛的智能算法。为了解决该算法应用过程中出现的滤波发散问题,文中运用协方差局地化方法,综合考虑先验地质模型的相关半径和观测数据的观测影响半径,计算临界半径长度,并在油藏模型模型中的水平和垂直方向引入局地化相关函数,滤除远距离观测数据产生的相关噪音,降低协方差矩阵计算过程中的伪相关。将改进的算法编程实现并运用理论实例进行验证,并对比反演得到的渗透率场与真实渗透率场,结果表明,改进后的理论在渗透率反演精度方面提高了28%,数据拟合速度提高了16%。反演的渗透率场能够清晰刻画出大孔道,对于优势通道识别、精细油藏描述具有重要意义。     

单一模型EnKF法更新油藏模型的探讨

近年来，人们开始利用集合卡尔曼滤波（EnKF）进行自动历史拟合，更新油藏模型。实验证明EnKF方法是一种有效的方法。EnKF方法要求一组初始实现，但是在实际应用中往往只有一个油藏或地质模型，这使得直接利用EnKF进行历史拟合，更新油藏模型比较困难。为此，提出了一种由一个模型派生出多个模型来满足EnKF基本要求的随机导航点方法，并试验在尽量少的模型条件下，利用EnKF更新油藏模型。模型实验表明，利用随机点抽样派生出多个模型，再利用EnKF可以实现单一模型更新，最终取得的模型与参考模型具有较好的一致性。     

基于流线EnKF油藏自动历史拟合

作为一种油藏参数动态估计的优化算法,将集合卡尔曼滤波EnKF与流线方法相结合,在算法实现过程中,选择具有更高计算精度和更快计算速度的流线模拟器进行预测,将油藏模型自动还原为一系列沿流线的一维模型,使数值弥散和受网格划分的影响达到最小,保持了明显的驱替前缘。采用序贯高斯协模拟方法生成一组地质模型,通过实时观测数据（产量及含水率等）,连续动态更新油藏模型的静态参数（渗透率及孔隙度等）、动态参数（压力及饱和度等）,同时实现流线分布的更新,直观地反映油藏流体在注采井之间的运动轨迹。采用Bayes理论阐述了流线EnKF数学原理,并通过拟合计算一个二维水驱油藏模型,验证了方法的有效性。     

Assisted history matching in the presence of frequent well intervention using generalize travel time inversion

In this paper we propose an efficient and novel technique that handles production discontinuities through a resampling of the production data, eliminating high frequency production details in a transformed domain. The technique also reduces non-monotonic behavior and results in a response more suitable for the GTTI based misfit calculations. Our proposed approach has been applied to an offshore turbidite reservoir with extensive well intervention resulting in highly detailed production responses. The static model contains more than three-hundred-thousand cells, complex sand depositional distribution combined with fault structures, four pairs of injector-producers, deviated producing wells and more than 8 years of production history. Previous history matching attempts using traditional approaches had difficulties matching production response at the individual well level. With our proposed modifications to the GTTI approach, a significant improvement was obtained on the well level match quality. Most importantly, by visualizing the streamlines and the dynamic adjustment of flow paths during history matching, we could easily identify the areas of inconsistency between the geologic model and the production data. The calibrated geologic model and streamline trajectories provided important insight about communication within sand channels, differences in flow paths and barriers that have not been included in the previous geologic and seismic interpretation.     

An Automatic History Matching Module with Distributed and Parallel Computing

Abstract

The data assimilation process of adjusting variables in a reservoir simulation model to honor observations of field data is known as history matching and has been extensively studied for a few decades. However, limited success has been achieved due to the high complexity of the problem and the large computational effort required in the real fields. Successful applications of the ensemble Kalman filter (EnKF) to reservoir history matching have been reported in various publications. The EnKF is a sequential method: once new data are available, only these data are used to update all the unknown reservoir properties while previous geological information is unused directly. In this method, multiple reservoir models rather than one single model are implemented, and each model is called a member. Conventionally, the impact of each member on the updating is equally treated. Another approach is the weighted EnKF. During the updating, the method weighs the contribution of each member through the comparison between the simulation response and the measurements. Better matching performance has been found in the weighted EnKF than in the conventional EnKF. To improve computational efficiency, two-level high-performance computing for reservoir history matching process is implemented in this research, distributing ensemble members simultaneously while simulating each member in a parallel style.

An automatic history-matching module based on the weighted EnKF and high-performance computing is developed and validated through a synthetic case operating from primary, waterflooding to flooding of water alternating with gas. The study shows that the weighted EnKF improves the matching results, and the high-performance computing process significantly reduces the history matching execution time.     

A framework to integrate history matching and geostatistical modeling using genetic algorithm and direct search methods

History matching is an inverse problem where the reservoir model is modified in order to reproduce field observed data. Traditional history matching processes are executed separately from the geological and geostatistical modeling stage due to the complexity of each area. Changes made directly on the reservoir properties generally yield inconsistent geological models. This work presents a framework to integrate geostatistical modeling and history matching process, where geostatistical images are treated as matching parameters. The traditional optimization methods normally applied in history matching generally use gradient information. The treatment of geostatistical images as matching parameters is difficult for these methods due to the strong non-linearities in the solution space. Therefore, another objective of this work is to investigate the application of two optimization methods: genetic algorithm and direct search method in the proposed framework. In order to accelerate the optimization process, two additional techniques are used: upscaling and distributed computing. Results are presented showing the viability of the genetic algorithm in the type of problem addressed in this work and also that direct search method can be used with some restriction. Finally, the benefits of distributed computing and the consistence of the upscaling process are shown.