The storage group has been studying the complications involved in geological sequestration of carbon dioxide, including the potential for leakage and assessment of environmental consequences. Their work has led to group members’ participation in IPCC reports on impacts on sequestration from manmade features, including abandoned wells.
Work on predicting the fate of injected CO2 is being pursued on a variety of fronts. While work continues on the in-house code Dynaflow, the team has also moved into working with analytical models that are considerably faster than complex simulators.
Since the beginning of the CMI grant, Jean-Hérve Prévost and colleagues have expanded the capabilities of Dynaflow to include multiphase, multicomponent flow. The model is now capable of simulating the behavior of CO2 and other components in brine systems, and initial simulations show significant improvement over previous models with less sophisticated treatment of chemistry and geomechanics. The team is currently adding calculation of pH to the model’s capabilities, which, along with subsequent geochemical routines, will allow study of the long-term fate of CO2.
On timescales of a few hundred years, most of the CO2 injected into a saline aquifer will not dissolve into the surrounding brine but instead remain as a separate, buoyant fluid. These early times are thus the critical period for potential leakage out of the injection reservoir. By focusing on single and two-fluid flow dynamics, Mike Celia’s group has developed new analytical models for CO2 plume evolution, including possible leakage through abandoned wells. These analytical solutions are much faster than full numerical simulators for estimating short-term reservoir leakage.
A study of the Alberta Basin indicates that this speed advantage will be important. Spatial analysis of existing wells in the basin indicates that they are the most significant potential leakage pathways for CO2. Because of the wells’ clustered distribution, the CO2 plume from a typical injection well could impinge upon hundreds of other wells. Estimating total reservoir leakage therefore requires modeling large numbers of wells simultaneously, which would be very time-consuming with more complex numerical models.
The group is currently working to expand the capabilities of the analytical model to include an arbitrary number of potentially leaky wells and multiple aquifers within the domain. These analytical solutions will then be combined with spatial analysis of well locations and cement degradation data (see below) to provide field-scale leakage estimates.
Deterioration of Cements
George Scherer’s group is studying degradation of well cements, which has the potential to increase reservoir CO2 leakage rates. Interaction with low pH groundwater could lead to rapid dissolution of 20% of the material in cements followed by structural collapse, but their behavior under sequestration conditions is not well known. After setting up a laboratory and making initial analyses to characterize sample cements, the group currently has experiments underway.
The experiments will focus on the physical and chemical changes that take place at the wall of a well where cement and rock meet. Tubular samples of a representative sandstone and limestone have been filled with relatively impermeable cement and will be exposed to brine solutions with varying temperatures and pH values for a year. At frequent intervals during the experiment, the group will examine the rock-cement interface for physical and chemical changes that would affect cement permeability.
As cement permeabilities are very low in relation to formation stones, changes in well cements are unlikely to result in large changes in CO2 transport through an aquifer. Compared to cap rocks with very low permeability, however, degradation of cement could make a well a more significant conduit of CO2. Future plans are to carry out similar experiments with cap rock samples supplied by BP.
Results of the corrosion experiments will provide input for the modeling studies by Prévost, including rates of dissolution of cement by flowing brine and cracking at the cement/formation interface from rising injection pressure.
Deep Aquifer Geochemistry
Catherine Peters and colleagues have been carrying out geochemical experiments to mimic the effects of CO2 injection under the temperature and pressure conditions found in deep aquifers.
Prior theoretical work had predicted that when CO2 is injected into a reservoir, metals released from silicate minerals could react with carbon in the aquifer to form new carbonate minerals that could sequester CO2. Calculations indicated that these reactions could take up 10-20% of the carbon injected into aquifers, but the speed of the reaction in a deep aquifer under high pressure was previously unknown.
A surprising result of the group’s experiments has been that although dissolution of precursor silicate minerals was relatively quick, solutions had to be highly supersaturated before new carbonate minerals would form. This indicates that sequestration via carbonate mineral formation may be difficult to achieve in real aquifers.
The group’s findings also indicate that, although temperature and pH have strong influences on dissolution rates, pressure has only a minor effect. Models can thus scale up existing rate laws for lower pressure conditions, making simulation of aquifers less complicated.
Finally, the group is using a computational approach to translate small-scale findings to the rock scale. For the first time, the group has used pore scale network models to simulate the effects of surface area, connectivity, and mineral heterogeneity on geochemical reaction rates.
Peter Jaffé and colleagues have been working on the impacts of leaks of CO2 from deep aquifers and their possible environmental risks.
Impacts of CO2 Leakage on Drinking Water
The team’s early work involved identifying contaminants that might be released if minerals in shallow aquifers were dissolved by CO2-rich fluids. Initial analysis of a USGS dataset indicated that arsenic was the only common element likely to exceed recommended maximum concentrations. However, as arsenic also becomes less mobile with increasing acidity, the group found that leakage of CO2 into shallow aquifers is not likely to impact water quality adversely.
Iron as a CO2 “Alarm”
Large CO2 fluxes into shallow soils would make soils anoxic and could result in high iron levels in groundwater. The appearance of this iron in local streams could thus serve as sort of CO2 “alarm.” To test the viability of this indicator, the group is performing soil column experiments to quantify the amount of iron that might be released from anoxic soils. This work also has implications for the sequestration of CO2 in soils as iron carbonate.
Risks from large point sources of CO2
In another program, field observations, combined with modeling of CO2 transport in soils, are helping to quantify risks from large underground point sources of carbon dioxide.
Satish Myneni leads an effort to study the impacts of high CO2 concentrations on soil chemistry and plant growth. Since the 1990’s, large amounts of carbon dioxide have leaked up from an active underground magma chamber beneath the Mammoth Mountain resort area in California, causing high concentrations of CO2 in local soils and the die-off of large numbers of trees.
Myneni’s team investigated the soil chemistry in regions with high CO2 fluxes. They found that minerals are more intensely weathered in these areas and that the nutrients released from this weathering were washed away rather than retained in soils. In addition, root systems of certain trees were altered by high CO2 concentrations in soils, spreading out at shallow depths rather than growing down into deep layers as they normally would.
To gain more insight into the impacts of high CO2, the group is examining the feasibility of installing instrumentation at Mammoth Mountain to continuously record CO2 concentrations throughout a soil profile. Current measurements are made only at the ground surface and provide no information about CO2 levels deeper within soils where impacts are observed.
Although Mammoth Mountain provides an excellent natural laboratory for studying a large CO2 point source, the young, ash-based soils around the resort are not a good analogue for average soils. Myneni and colleagues are thus carrying out complementary work on soils influenced by another kind of CO2 point source – underground coal fires. Coal fires in Pennsylvania cause large fluxes of carbon dioxide into overlying soils that are more typical in their makeup than those at Mammoth Mountain. Comparing the impact of high CO2 in the two soil types will therefore allow the researchers to predict whether impacts seen at Mammoth Mountain can be extrapolated to other areas that experience high CO2 fluxes.