Department of Civil & Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana, Illinois.
Department of Aerospace & Mechanical Engineering, University of Notre Dame, Notre Dame, Indiana, USA.
J Contam Hydrol. 2018 May;212:14-27. doi: 10.1016/j.jconhyd.2017.09.005. Epub 2017 Sep 23.
We employed the color-fluid lattice Boltzmann multiphase model to simulate liquid CO displacing water documented in experiments in a 2D heterogeneous micromodel at reservoir pressure conditions. The main purpose is to investigate whether lattice Boltzmann simulation can reproduce the CO invasion patterns observed in these experiments for a range of capillary numbers. Although the viscosity ratio used in the simulation matches the experimental conditions, the viscosity of the fluids in the simulation is higher than that of the actual fluids used in the experiments. Doing so is required to enhance numerical stability, and is a common strategy employed in the literature when using the lattice Boltzmann method to simulate CO displacing water. The simulations reproduce qualitatively similar trends of changes in invasion patterns as the capillary number is increased. However, the development of secondary CO pathways, a key feature of the invasion patterns in the simulations and experiments, is found to occur at a much higher capillary number in the simulations compared with the experiments. Additional numerical simulations were conducted to investigate the effect of the absolute value of viscosity on the invasion patterns while maintaining the viscosity ratio and capillary number fixed. These results indicate that the use of a high viscosity (which significantly reduces the inertial effect in the simulations) suppresses the development of secondary CO pathways, leading to a different fluid distribution compared with corresponding experiments at the same capillary number. Therefore, inertial effects are not negligible in drainage process with liquid CO and water despite the low Reynolds number based on the average velocity, as the local velocity can be much higher due to Haines jump events. These higher velocities, coupled with the low viscosity of CO, further amplifies the inertial effect. Therefore, we conclude that caution should be taken when using proxy fluids that only rely on the capillary number and viscosity ratio in both experiment and simulation.
我们采用彩色流格子玻尔兹曼多相模型，模拟了在储层压力条件下二维非均质微模型中实验记录的液体 CO 驱替水的过程。主要目的是研究格子玻尔兹曼模拟是否可以再现这些实验中观察到的 CO 侵入模式，以及在一系列毛细数范围内是否可以再现。尽管模拟中使用的粘度比与实验条件相匹配，但模拟中流体的粘度高于实验中实际使用的流体。这样做是为了提高数值稳定性，这是在使用格子玻尔兹曼方法模拟 CO 驱替水时文献中常用的策略。模拟再现了随着毛细数增加，侵入模式变化的定性相似趋势。然而，在模拟中，与实验相比，二次 CO 途径的发展，即模拟和实验中侵入模式的一个关键特征，发生在更高的毛细数。进行了额外的数值模拟，以研究在保持粘度比和毛细数固定的情况下，粘度绝对值对侵入模式的影响。这些结果表明，高粘度的使用（这大大降低了模拟中的惯性效应）抑制了二次 CO 途径的发展，导致在相同毛细数下与相应实验相比，流体分布不同。因此，尽管基于平均速度的雷诺数较低，但在液体 CO 和水的排水过程中，惯性效应不可忽略，因为由于 Haines 跳跃事件，局部速度可能会高得多。这些更高的速度，加上 CO 的低粘度，进一步放大了惯性效应。因此，我们得出结论，在实验和模拟中仅依赖毛细数和粘度比使用替代流体时应谨慎。
