Semi-analytical models of single injection plumes
During the last year Michael Celia’s group continued work on large-scale modeling of leakage along existing wells, and the large-scale leakage model they have developed is now being included in a comprehensive risk-assessment tool being developed by Los Alamos National Laboratory. In this general area, the researchers have continued to develop and test the semianalytical solutions for CO2 and brine migration, including leakage along wells. The current state of this model is that they now have put together almost all of the key components, including: the CO2 plume model, the leakage model for multiple wells and multiple aquifers, and the local upconing model (which can have a strong impact on leakage behavior). The only major component missing from the model is a detailed phase-change package for rapid leaks along wells. Celia and colleagues are debating how much detail is needed, and have not yet reached a consensus on this issue.
The group is continuing to use the Alberta Basin as source for hypothetical test cases. The Wabamun Lake region, southwest of Edmonton, has an extensive data base put together by their continuing collaborator, Stefan Bachu, and his team in Edmonton. Those data are being used as a basis for simulations, with probability distributions for leakage being the main output. The team has also considered vertical distributions of wells, and is in the process of evaluating risk reduction as a function of depth of injection. They have also begun to examine injectivity as a function of depth, as part of the overall study of depth of injection.
Basin-scale injection model
In addition to these semi-analytical solutions, which have been applied at the scale of a single injection plume (of order 1,000 km2), Celia’s group has also put together a new modeling approach that they believe is appropriate for modeling at length scales much larger than a single plume. They do not have any publications on this yet, but the general idea is to relax some of the assumptions required for analytical or semi-analytical solutions, while maintaining sufficient simplicity to achieve reasonably efficient numerical solutions. The researchers plan to use this approach to model basin-wide systems, with a focus on overall CO2 injection volumes and locations, and the associated migration of brine within and across formations. Such a basin-wide model would allow for direct interaction with models of surface facilities, can consider optimization for a range of objective functions, and can answer questions about short- and long-term migration of displaced brines.