Benchmarking of vertically-integrated CO2 flow simulations at the Sleipner Field, North Sea
Benchmarking of vertically-integrated CO2 flow simulations at the Sleipner Field, North Sea
复制标题
北海斯莱普纳油田垂直整合二氧化碳流模拟的基准测试
DOI：
10.1016/j.epsl.2018.03.038
复制
发表时间：
2018
影响因子：
5.3
通讯作者：
R. Chadwick
中科院分区：
文献类型：
--
作者：
L. Cowton;J. Neufeld;N. White;M. Bickle;G. Williams;J. White;R. Chadwick
数值模拟在识别和评估可能适合未来碳捕获和封存项目的地下储层方面发挥着至关重要的作用。流量模拟的准确性进行测试基准对历史观测正在进行的CO2注入网站。在位于北海的Sleipner项目中，一套时间推移地震反射勘测使CO2在储层顶部的三维分布能够确定为时间的函数。以前的尝试已经使用达西流动模拟器来模拟CO2在整个这一层的迁移，给定注入量随时间的变化和注入点的位置。主要是由于计算的限制，防止充分探索模型参数空间，这些模拟通常无法匹配观测到的CO2分布作为空间和时间的函数。为了规避这些限制，我们开发了一个垂直集成的流体流动模拟器，是基于地形控制，多孔重力流的理论。这种计算效率高的方案可用于反演所需的储层渗透率的空间分布，以尽量减少观测和计算的CO2分布之间的差异。当假设储层渗透率均匀时，反演模型无法充分匹配CO2在储层顶部的迁移。然而，如果允许绘制的通道存款的宽度和渗透率独立变化，则获得了观测和计算的CO2分布之间令人满意的匹配。最后，该算法的能力，预测CO2在水库顶部的流量进行了评估。通过将完整的地震反射勘测集划分为训练子集和验证子集，我们发现与训练子集匹配所需的渗透率空间模式可以成功地预测验证子集的CO2迁移。这种能力表明，有一定把握地预测未来的移徙模式是可行的。然而，我们的分析强调了在没有额外观测约束的情况下，估计远离CO2波及区域的储层参数的困难。
Numerical modeling plays an essential role in both identifying and assessing sub-surface reservoirs that might be suitable for future carbon capture and storage projects. Accuracy of flow simulations is tested by benchmarking against historic observations from on-going CO2injection sites. At the Sleipner project located in the North Sea, a suite of time-lapse seismic reflection surveys enables the three-dimensional distribution of CO2at the top of the reservoir to be determined as a function of time. Previous attempts have used Darcy flow simulators to model CO2migration throughout this layer, given the volume of injection with time and the location of the injection point. Due primarily to computational limitations preventing adequate exploration of model parameter space, these simulations usually fail to match the observed distribution of CO2as a function of space and time. To circumvent these limitations, we develop a vertically-integrated fluid flow simulator that is based upon the theory of topographically controlled, porous gravity currents. This computationally efficient scheme can be used to invert for the spatial distribution of reservoir permeability required to minimize differences between the observed and calculated CO2distributions. When a uniform reservoir permeability is assumed, inverse modeling is unable to adequately match the migration of CO2at the top of the reservoir. If, however, the width and permeability of a mapped channel deposit are allowed to independently vary, a satisfactory match between the observed and calculated CO2distributions is obtained. Finally, the ability of this algorithm to forecast the flow of CO2at the top of the reservoir is assessed. By dividing the complete set of seismic reflection surveys into training and validation subsets, we find that the spatial pattern of permeability required to match the training subset can successfully predict CO2migration for the validation subset. This ability suggests that it might be feasible to forecast migration patterns into the future with a degree of confidence. Nevertheless, our analysis highlights the difficulty in estimating reservoir parameters away from the region swept by CO2without additional observational constraints.