The Storage group will continue work to characterize corroded cements, expand field studies, refine and apply the Dynaflow and analytical-numerical models, and develop a hierarchical modeling framework for modeling basin-scale injection scenarios.
Underground Facility for Leakage Testing
A proposal has been submitted to NSF (with Catherine Peters as the lead investigator and Jean Prévost and George Scherer among the PIs) to build a facility for testing of leakage on a large scale. A deep underground science and engineering laboratory (DUSEL) is being constructed in an abandoned gold mine in South Dakota, with the primary goal of carrying out particle physics experiments. Proposals have also been solicited for use of the facility to do experiments in other fields, such as biology and geomechanics. The researchers propose to use the long vertical shafts to create a pressure vessel with a length comparable to the height of the Empire State Building to simulate leakage from a reservoir. The scale would enable following phase changes as supercritical CO2 expands, and measurement of the change in leakage rate through cement with controlled cracks or annular gaps. This would provide a direct test of the group’s simulations of leakage through wells. The proposal was developed in collaboration with colleagues at the Lawrence Berkeley National Laboratory.
Field Studies of Well Properties
In terms of applications, Michael Celia’s group will continue to work with Walter Crow, Brian Williams, and others at BP to expand the number of wellbore tests and the associated database of effective permeability values for cemented annular regions of old wells. They will also continue to work with Stefan Bachu and Teresa Watson to further integrate their scoring system, based on ‘soft’ well data, into probabilistic models for leakage along wells. Finally, they will begin to develop a new field test site that builds on initial work at the Wabamun Lake area (Figure 10 and Figure 11). Because it seems likely that several CCS operations will be developed around the greater Edmonton area, in association with oil sands production, the researchers are interested in how multiple large-scale CCS operations will interact and interfere with one another. That requires modeling on a scale much larger than the domain of Figure 10, which covers 2,500 km2 contains roughly 1,300 old wells. If the domain size were expanded by an order of magnitude, the number of wells should increase in proportion, and the overall computational challenge to increase by more than a linear multiple. The team plans to initiate data collection activities for this much expanded domain over the next year, and to develop a hierarchical modeling framework with these kinds of applications in mind. Geological and geometric heterogeneities, fault zones, interacting plumes, and potential leakage along perhaps tens of thousands of wells will all need to be included in the modeling. The team sees this as their next large challenge, one that will allow subsurface models to move toward an integrated assessment of CCS operations on a regional scale.
Evaluating Cement Corrosion and Impacts
In the coming year, the Scherer-Prévost group will continue to focus on their two priorities: (1) characterizing the mechanical and transport properties of corroded cement, and (2) developing the software tools needed to model leakage of CO2. The modeling work is currently directed toward perfecting the flash calculation, to permit prediction of phase changes as the fluid rises through zones of decreasing temperature and pressure. This research is being performed by Prof. Jean Prévost in collaboration with Dr. Lee Y. Chin (ConocoPhillips). They are still hoping to have meaningful collaboration with Bruno Huet at Schlumberger, but that has not been possible to date. In the past year, the experimental work by Ed Matteo has been largely devoted to developing experimental methods, which are now being put to use. In the near future they expect to quantify the diffusion coefficient in cements with various degrees of corrosion damage, and to determine the indentation strength of those materials. These data are essential inputs for the modeling of leakage.
Large-Scale Injection Simulations
Celia and colleagues plan to continue to develop their semi-analytical and numerical sharpinterface models to allow for more complex geometry, geology, and fluid behavior, while still maintaining computational efficiencies. This will include development of analytical solutions that include diffuse leakage across caprock formations. While they do not believe this is important in terms of mass transport, it can be quite important when analyzing the pressure response to injection systems. The pressure response, in turn, is important when identifying the “Area of Review”, an important quantity in the EPA guidelines. Their approach will be to take advantage of capillary exclusion for the CO2 but to allow brine to leak across caprock formations. This reduces the problem to one of single-phase flow with moving boundaries, for which some analytical solutions can be derived. The researchers will integrate these solutions, with varying degrees of complexity, into their overall semi-analytical model for injection and leakage.
The group will also expand its numerical sharp-interface model to include more complex heterogeneity, to allow for sub-scale representation of faults as well as leaky wells, and to integrate more complex phase behavior along concentrated leakage pathways. The overall approach is to solve the sharp-interface equations on coarse grids within each layer, and to use local analytical solutions to capture sub-grid-scale behavior like local upconing around a well or a fault, with concomitant Peaceman-type corrections for the local pressure field. This kind of approach represents a multi-scale hybrid numerical-analytical solution, with numerical solutions used at the large scale and analytical corrections used on the fine scale. The researchers plan to extend this concept to develop a more general ‘hierarchical’ modeling framework, where they use large-scale semi-analytical solutions in parts of the overall domain where they are justified, numerical sharp-interface models where heterogeneity and other factors require such solutions, and local analytical corrections to model the effects of concentrated leakage pathways. While simple in concept, there are a number of computational issues to be worked out, and they will begin to develop the details of this framework over the next year.