Factors determining rapid and efficient geologic storage of CO₂
Implementing geological carbon sequestration at a scale large enough to mitigate emissions will involve the injection of supercritical CO₂ into deep saline aquifers. The principal technical risks associated with such injection are that (i) buoyant CO₂ will migrate out of the storage formation; (ii) pressure elevation during injection will limit storage rates and/or fracture the storage formation; and (iii) groundwater resources will be contaminated, directly or indirectly, by brine displaced from the storage formation. An alternative to injecting CO₂ as a buoyant phase is to dissolve it into brine extracted from the storage formation, then inject the CO₂-saturated brine into the storage formation. This "surface dissolution" strategy completely eliminates the risk of buoyant migration of stored CO₂. It greatly mitigates the extent of pressure elevation during injection. It nearly eliminates the displacement of brine. To gain these benefits, however, it is essential to determine the costs of this method of risk reduction. This work provides a framework for optimization of the process, and hence for cost minimization. Several investigations have tabulated the storage capacity for CO₂ in regions around the world, and it is widely accepted that sufficient pore volume exists in deep subsurface formations to permit large-scale sequestration of anthropogenic CO₂. Given the urgency of implementing geologic sequestration and other emissions-mitigating technologies (storage rates of order 1 Gt C per year are needed within a few decades), the time required to fill a target formation with CO₂ is just as important as the pore volume of that formation. To account for both these practical constraints we describe in this work a time-weighted storage capacity. This modified capacity integrates over time the maximum injection rate into a formation. The injection rate is a nonlinear function of time, formation properties and boundary conditions. The boundary conditions include the maximum allowable injection pressure and the nature of the storage formation (closed, infinite-acting, constant far-field pressure, etc.) The time-weighted storage capacity approaches the volumetric capacity as time increases. For short time intervals, however, the time-weighted storage capacity may be much less than the volumetric capacity. This work describes a method to compute time-weighted storage capacity for a database of more than 1200 North American oil reservoirs. Because all of these reservoirs have been commercially developed, their formation properties can be regarded as representative of aquifers that would be attractive targets for CO₂ storage. We take the product of permeability and thickness as a measure of injectivity for a reservoir, and the product of average areal extent, net thickness and porosity as a measure of pore volume available for storage. We find that injectivity is not distributed uniformly with volume: the set of reservoirs with better than average injectivity comprises only 10% of the total volumetric storage capacity. Consequently, time weighted capacity on time scale of a few decades is 10% to 20% of the nominal volumetric capacity. The non-uniform distribution of injectivity and pore volume in the database coupled with multiphase flow effects yields a wide distribution of “filling times”, i.e. the time required to place CO₂ up to the boundaries of the formation. We define two limiting strategies based on fill times of the storage structures in the database and use them to calculate resource usage for a target storage rate. Since fill times are directly proportional to injectivity, smallest fill time corresponds to best injectivity and largest fill time corresponds to smallest injectivity. If best injectivity structures are used first, then the rate at which new structures would be needed is greater than if worst injectivity structures are used first. A target overall storage rate could be maintained for longer period of time when worst injectivity structures are used first. Because of the kh vs PV correlation, most of the pore volume remains unused when no extraction wells are used. Extraction wells require disposal of produced brine, which is a significant challenge, or beneficial use of the brine. An example of the latter is the surface dissolution process described in this thesis, which would enable use of a much greater fraction of the untouched pore volume.