Ninth Year Annual Report:
Carbon Storage: Large-Scale Modeling of CO2 Injection and Transport
The Celia Group develops and uses models on large scales suitable for risk-analysis of carbon storage in geological basins. This year, the group has made new strides in simulating the longterm fate of injected CO2, expanding the capabilities of existing models, and testing them against field data.
Long-Time Models for CO2 Fate and Transport
Until recently, the groupís large-spatial-scale models had focused on leakage analysis during the injection phase of an operation. For this time frame, the system is dominated by pressure drive from the injection operation and buoyant forces associated with the free-phase CO2. For longer post-injection time scales, capillary trapping of migrating CO2 is important, as is larger-scale dissolution of CO2 into the background brine phase, especially when convective mixing in the aqueous phase is included.
The group now has fairly general models for longer-time CO2 fate and transport including capillary trapping and convective mixing. These models are based on an assumption of vertical equilibrium and the associated vertical segregation of the buoyant fluid phases. Vertical integration is applied to take advantage of the stratified flows in the two-phase transport system. An upscaled version of the convective mixing equation is used in the overall transport model. Details of the model can be found in Gasda et al. (2010).
The researchers have applied these models to several test problems, with results demonstrating the importance of both dimensionality and the role of convective mixing in limiting the upslope migration of free-phase CO2. A field application using detailed date from the Johanson Formation (North Sea) shows how complex geology, including spatially variable permeability and porosity and complex formation geometry, can all be incorporated easily into the modeling framework. Celia and colleagues have simulated several different injection scenarios, and tracked through time the amount and location of CO2 in free phase, dissolved phase, and residually-trapped phase. Figure 10 shows a series of panels showing the spatial extent of the three forms of CO2 at different times and a plot showing the fraction of the injected CO2 residing in each of the three forms as a function of time.
Figure 11 may be seen as a quantification of the "storage security" plot in the IPCC (2005) report on CCS. This general numerical model is being integrated into the groupís overall modeling framework, which is described in the "future work" section below.
Hybrid and Multi-Scale Models
The Celia Group is continuing development of a general modeling framework within which they use a multi-scale approach to incorporate numerical and (semi-)analytical approaches to model CO2 injection, migration, and leakage. The researchers now have a robust set of semi-analytical modeling tools that can model multiple layers and multiple leaky wells. They also have a set of robust numerical simulators (like the one described in the previous section) that allow for more general problems to be solved. Within these numerical codes, local-scale analytical solutions allow wells to be represented without grid refinement around the wells. This had previously been done for leaky wells, and there is now a new formulation for injection wells that defines both the sub-grid-scale CO2 plume within the grid cell and the associated pressure distribution. The new formulation gives much improved results as compared to a more traditional Peacemantype correction.
All of these models are based on the assumption of vertical equilibrium with buoyant stratification and a sharp interface separating the two fluid phases. Recent work to include a capillary fringe in place of a sharp interface demonstrates the potential importance of local-scale capillarity in mitigating the updip migration of buoyant CO2. Capillarity tends to retard significantly the buoyant migration of the leading tip of free-phase CO2, leading to slower plume migration and a modified shape of the leading edge of the plume. This work on capillarity has been done by the groupís Norwegian partners Jan Nordbotten and Helge Dahle.
Applications to the Alberta Basin
Celia and colleagues continue to use the data set developed for the Wabamun Lake area to study a number of aspects related to leakage risk and old wells. In addition to the issues of injectivity limitations and risk of leakage versus depth of injection, they have continued studies to determine the circumstances under which a particular level of leakage will be reached. For example, if a bimodal lognormal distribution describes the probability distribution for permeability values along segments of old wells, then the two main parameters used to describe the distribution can be used to define regions within which leakage will not exceed a specified threshold (to 95% confidence).
Similar parametric studies have been carried out using parameters from the scoring system for old wells developed by Watson and Bachu (2008, 2009). The group has also performed initial studies focused on strategies for monitoring injection sites. Assuming leakage patterns that arise from flows along old wells, they have considered the required density of observation points for a direct monitoring network and the impact of using observations of actual CO2 plumes versus observations of pressure perturbations. Not surprisingly, monitoring is much more effective when based on pressure measurements because pressure pulses propagate much further in formations that have leakage and therefore can be detected much more easily.
In collaboration with the Capture Group, the Celia team has begun to explore active brine management associated with CCS operations and with injection operations. Regarding CCS surface operations, the increased water demands at power plants with CCS may make water supply a critical issue in arid or semi-arid regions. This in turn motivates the possible use of deep subsurface water in CCS plant operations. If this is pursued, then strategies for joint management of both CO2 and brine should be developed. The researchers have begun to study this problem through consideration of brine extraction scenarios, possible use of extraction wells to manage excessive pressure buildup, and possible use of extraction wells to allow for plume steering and mitigation of CO2 migration.
Interpretation of Borehole Experiments
The Celia group continues to work with Walter Crow and others at BP to improve analysis methods for data collected when old wells are reentered and Vertical Interference Tests are performed. The researchers currently have dedicated software to estimate the leakage and formation parameters, but it remains as a trial-and-error process. They plan to automate this procedure to enhance their abilities to estimate critical parameters for leakage and risk analysis.