A Study on Long-term Stability of CO2 Geological Sequestration

二氧化碳排放量的增加被认为是全球变暖的一个因素，人们研究了各种封存技术。然而，人们对二氧化碳的长期行为还没有很好的了解。注入地下含水层的二氧化碳以气相、液态或固相的形式稳定。溶解在液体中的二氧化碳在含水层中通过对流-弥散现象扩散。评价孔隙水平非均质性对渗流行为的影响是十分必要的。在本研究中，使用砂岩、碳酸盐岩和一包玻璃微珠进行了实验。流动模型采用对流弥散(CDE)模型和Coats-Smith模型，并与实验结果进行了拟合。在一组玻璃微珠和砂岩上成功地应用了ODE技术。此外，还证实了弥散系数与流速成正比，与样品的种类无关。另一方面，对于白云岩，CDE的应用并不成功。这是由于次生孔隙和孔隙层面的非均质性所致。当应用反映非均质性的Coats-Smith模型时，得到了较好的匹配结果。样品中存在大量的滞留部分，与不具有这种非均质性的岩石和玻璃微珠相比，具有孔隙级非均质性的岩石可能具有较大的弥散系数。当使用流动模拟器来模拟ODE扩散时，使用从Coats-Smith模型获得的扩散系数可能会导致错误的预测。要获得可接受的预测结果，它是正确的。有必要设置一个大于实际的表观系数

The increase in discharge of carbon dioxide is considered as a factor of global warming, and various sequestration technologies are studied. However, the long-term behavior of carbon dioxide is not well understood. The carbon dioxide injected into underground aquifer is stabilized as a gaseous phase, a liquid phase, or a solid phase. The carbon dioxide dissolved in the liquid phase spreads by convection-dispersion phenomena in the aquifer. It is necessary to evaluate the influence of pore-level heterogeneity on the flow behavior. In this research, the experiment using sandstone, carbonate rock, and a pack of glass bead was conducted. A convection-dispersion (CDE) model and the Coats-Smith model were adopted as a flow model, and matching with an experiment result was performed. Application of ODE was successful for a pack of glass bead and sandstone. Moreover, it has been confirmed that the dispersion coefficient was proportional to the flow velocity regardless of the kind of sample. On the other hand, application of CDE was not successful for the dolomite. This is because the secondary pore and the pore-level heterogeneity. When the Coats-Smith model reflecting the heterogeneity was applied, the good matching result was obtained. A large amount of stagnation portion exists in the sample, and it turned out that the rock which has the pore-level heterogeneity may have a large dispersion coefficient compared with the rock and the glass bead without such heterogeneity. When using flow simulators which models dispersion by ODE, using the dispersion coefficient obtained from the Coats-Smith model may result in erroneous predictions. To obtain acceptable-prediction results, it is. necessary to set up an apparent coefficient that is larger than the actual