Various approaches are used to evaluate the capacity of saline aquifers to store CO2, resulting in a wide range of capacity estimates for a given aquifer. The two approaches most used are the volumetric "open aquifer" and "closed aquifer" approaches. We present four full-scale aquifer cases, where CO2 storage capacity is evaluated both volumetrically (with "open" and/or "closed" approaches) and through flow modeling. These examples show that the "open aquifer" CO2 storage capacity estimation can strongly exceed the cumulative CO2 injection from the flow model, whereas the "closed aquifer" estimates are a closer approximation to the flow-model derived capacity. An analogy to oil recovery mechanisms is presented, where the primary oil recovery mechanism is compared to CO2 aquifer storage without producing formation water; and the secondary oil recovery mechanism (water flooding) is compared to CO2 aquifer storage performed simultaneously with extraction of water for pressure maintenance. This analogy supports the finding that the "closed aquifer" approach produces a better estimate of CO2 storage without water extraction, and highlights the need for any CO2 storage estimate to specify whether it is intended to represent CO2 storage capacity with or without water extraction.
Heterogeneities of diffusion properties are likely to influence the effective matrix diffusion coefficient determined from tracer breakthrough curves. The objectives of this study are (1) to examine if it is appropriate to use a single, effective matrix diffusion coefficient to predict breakthrough curves in a fractured formation, (2) to examine if a postulated scale dependence of the effective matrix diffusion coefficient is caused by heterogeneity in diffusion properties, and (3) to examine whether multirate diffusion results in the previously observed time dependence of the effective matrix diffusion coefficient. The results show that the use of a single effective matrix diffusion coefficient is appropriate only if the interchannel and intrachannel variability of diffusion properties is small. The scale dependence of the effective matrix diffusion coefficient is not caused by the studied types of heterogeneity. Finally, the multirate diffusion process does not result in the time dependence of the effective matrix diffusion coefficient.
