A quantitative methodology to assess the risks to human health from CO 2 leakage into groundwater Erica R. Siirila a,c,⇑ , Alexis K. Navarre-Sitchler b , Reed M. Maxwell a,c , John E. McCray b,c a Department of Geology and Geological Engineering, Colorado School of Mines, Golden, CO 80401, United States b Environmental Science and Engineering Division, Colorado School of Mines, Golden, CO 80401, United States c Hydrologic Science and Engineering Program, Colorado School of Mines, Golden, CO 80401, United States article info Article history: Available online 1 December 2010 Keywords: Carbon Capture and Storage Metal mobilization Carbon dioxide leakage Monte Carlo Human health risk Joint Uncertainty and Variability abstract Leakage of CO 2 and associated gases into overlying aquifers as a result of geologic carbon capture and sequestration may have adverse impacts on aquifer drinking-water quality. Gas or aqueous-phase leak- age may occur due to transport via faults and fractures, through faulty well bores, or through leaky con- fining materials. Contaminants of concern include aqueous salts and dissolved solids, gaseous or aqueous-phase organic contaminants, and acidic gas or aqueous-phase fluids that can liberate metals from aquifer minerals. Here we present a quantitative risk assessment framework to predict potential human health risk from CO 2 leakage into drinking water aquifers. This framework incorporates the potential release of CO 2 into the drinking water aquifer; mobilization of metals due to a decrease in pH; transport of these metals down gradient to municipal receptors; distributions of contaminated groundwater to multiple households; and exposure and health risk to individuals using this water for household purposes. Additionally, this framework is stochastic, incorporates detailed variations in geo- logical and geostatistical parameters and discriminates between uncertain and variable parameters using a two-stage, or nested, Monte Carlo approach. This approach is demonstrated using example simulations with hypothetical, yet realistic, aquifer characteristics and leakage scenarios. These example simulations show a greater risk for arsenic than for lead for both cancer and non-cancer endpoints, an unexpected finding. Higher background groundwater gradients also yield higher risk. The overall risk and the associ- ated uncertainty are sensitive to the extent of aquifer stratification and the degree of local-scale disper- sion. These results all highlight the importance of hydrologic modeling in risk assessment. A linear relationship between carcinogenic and noncarcinogenic risk was found for arsenic and suggests action levels for carcinogenic risk will be exceeded in exposure situations before noncarcinogenic action levels, a reflection of the ratio of cancer and non-cancer toxicity values. Finally, implications for ranking aquifer vulnerability due to geologic configuration, aquifer mineralogy, and leakage scenarios are discussed. Ó 2010 Elsevier Ltd. All rights reserved. 1. Introduction Although geologic Carbon Capture and Storage (CCS) has re- cently become a viable option to reduce atmospheric CO 2 emis- sions, a quantitative methodology to assess the risks to human health from CO 2 leakage has yet to be developed. One of the ben- efits of large-scale CCS implementation is the large volume of CO 2 potentially sequestered, spanning as large as 5–6 km in radial extent [1,2]. The expansive area covered, however, causes concern for a higher probability of leakage pathways to be present [3]. Be- cause CO 2 gas is buoyant, if leakage pathways do exist, the gas may flow upwards towards potable drinking water resources. Such pathways include, but are not limited to (1) direct leaks such as abandoned wellbores, faults or fractures and (2) diffuse leaks such as permeable caprock material and micro-fractures in the caprock, e.g. [4–7]. If CO 2 leakage was to occur, one of the primary concerns is the contamination of potable water via the mobilization of toxic metals from aquifer material induced through a drop in groundwa- ter pH, e.g. [8–12]. Metal contamination is possible if the overlying aquifer material is metal bearing, and if CO 2 leakage occurs. The framework presented here quantifies the baseline human health risks associated with a leakage of CO 2 by examining multi- ple pathways of exposure through a probabilistic risk assessment given that leakage has occurred. Other studies examine risk of sequestration failure, e.g. [13], the probability and degree to which a leak occurs, e.g. [14–17], and probability of high volume (10 3 kg to 10 4 kg CO 2 ) surface vents from abandoned wells, e.g. [18]. These studies are important in creating a comprehensive understanding 0309-1708/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.advwatres.2010.11.005 ⇑ Corresponding author at: Hydrologic Science and Engineering Program, Colo- rado School of Mines, Golden, CO 80401, United States. E-mail address: esiirila@mymail.mines.edu (E.R. Siirila). Advances in Water Resources 36 (2012) 146–164 Contents lists available at ScienceDirect Advances in Water Resources journal homepage: www.elsevier.com/locate/advwatres