Detecting potential impacts of deep subsurface CO2 injection on shallow drinking water

Presented here are results from one aspect of collective research conducted at Gulf Coast Carbon Center, BEG, Jackson School at UT Austin. The biggest hurdle to public acceptance of CCS is to show that drinking water resources will not be impacted. Since late 1990s our group has been supported by US DOE NETL and private industry to research how best to detect potential impacts to shallow (0 to ~0.25 km) subsurface drinking water from deep (~1 to 3.5 km) injection of CO2. Work has and continues to include (1) field sampling and testing, (2) laboratory batch experiments, (3) geochemical modeling. The objective has been to identify the most sensitive geochemical indicators using data from research-level investigations, which can be economically applied on an industrial-scale. The worst-case scenario would be introduction of CO2 directly into drinking water from a leaking wellbore at a brownfield site. This is unlikely for a properly screened and/or maintained site, but needs to be considered. Our results show aquifer matrix (carbonate vs. clastic) to be critical to interpretation of pH and carbonate (DIC, Alkalinity, and δ13C of DIC) parameters because of the influence of water-rock reaction (buffering vs. non-buffering) on aqueous geochemistry. Field groundwater sampling sites to date are Cranfield, MS and SACROC, TX CO2-EOR oilfields. Two major aquifer types are represented, one dominated by silicate (Cranfield) and the other by carbonate (SACROC) water-rock reactions. We tested sensitivity of geochemical indicators (pH, DIC, Alkalinity, and δ13C of DIC) by modeling the effects of increasing pCO2 on aqueous geochemistry, and laboratory batch experiments, both with partial pressure of CO2 gas (pCO2) at 1x105 Pa (1 atm). Aquifer matrix and groundwater data provided constraints for the geochemical models. We used results from modeling and batch experiments to rank geochemical parameter sensitivity to increased pCO2 into weakly, mildly and strongly sensitive categories for both aquifer systems. DIC concentration is strongly sensitive to increased pCO2 for both aquifers; however, CO2 outgassing during sampling complicates direct field measurement of DIC. Interpretation of data from in-situ push-pull aquifer tests is ongoing and will be used to augment results summarized here. We are currently designing groundwater monitoring plans for two additional industrial-scale sites where we will further test the sensitivity and utility of our sampling approach.