Effects on Groundwater Quality of CO2 Leakage from Geological Storage

An experiment simulating the effect of CO2 leakage from geological storage on shallow potable aquifers was physically simulated in a shallow potable aquifer at Vrøgum plantation Western Denmark by injection of 1600 kg of gas phase CO2 over 72 days at 10 and 5 meters below surface. The groundwater chemistry was monitored by sampling of groundwater from a network of samplers, giving a detailed description of the development in the groundwater chemistry in time and space. A 1D-dimensional reactive transport model was constructed based on field and laboratory results and subsequently used to explore the contributions of various geochemical processes to explain observed results. It is proposed that dissolution of finite amounts of gibbsite releases Al3+, which drives cation exchange. This main process is able to explain the majority of water chemistry changes observed at Vrøgum including: a pulse like effect showing a fast peak and return towards background levels for alkalinity and dissolved ion concentrations; and increasing and persistent acidification via buffering exhaustion. Model processes were confirmed by simulation of a batch experiment conducted on the Vrøgum glacial sand, employing the same processes and sediment parameters. The reactive transport model was subsequently used to make long term predictions and explore various scenarios. Long term predictions, not surprisingly, suggest the pulse of elevated ions travels with advective flow succeeded by a zone of increasing acidification. Model runs at higher PCO2 (implying greater depths) suggest amplification of effects, i.e. greater peaks and more rapid and severe acidification. Calcite limits acidification, however, induces additional Ca2+ driven ion exchange, leading to more significant changes in chemistry. Although the model is site specific, the reactive minerals are common and the results should indicate general risks posed to water resources from CCS leakage and the parameters that should be in focus in monitoring programs.