Almost all studies of the potential for well leakage have been laboratory based, as there are limited data on field-scale leakage. Laboratory experiments show that CO2 and CO2-saturated brine still react with cement and casing when leakage occurs by diffusion only. The rate of degradation, however, is transport-limited and alteration of cement and casing properties is low. When a leakage path is already present due to cement shrinkage or fracturing, gaps along interfaces (e.g. casing/cement or cement/rock), or casing failures, chemical and mechanical alteration have the potential to decrease or increase leakage risks. Laboratory experiments and numerical simulations have shown that mineral precipitation or closure of strain-induced fractures can seal a leak pathway over time or conversely open pathways depending on flow-rate, chemistry, and the stress state. Experiments with steel/cement and cement/rock interfaces have indicated that protective mechanisms such as metal passivation, chemical alteration, mechanical deformation, and pore clogging can also help mitigate leakage. The specific rate and nature of alteration depend on the cement, brine, and injected fluid compositions. For example, the presence of co-injected gases (e.g. O2, H2S, and SO2) and pozzolan amendments (fly ash) to cement influences the rate and the nature of cement reactions. A more complete understanding of the coupled physical–chemical mechanisms involved with sealing and opening of leakage pathways is needed.
An important challenge is to take empirically based chemical, mechanical, and transport models reviewed here and assess leakage risk for carbon storage at the field scale. Field observations that accompany laboratory and modeling studies are critical to validating understanding of leakage risk. Long-term risk at the field scale is an area of active research made difficult by the large variability of material types (cement, geologic material, casing), field conditions (pressure, temperature, gradient in potential, residence time), and leaking fluid composition (CO2, co-injected gases, brine). Of particular interest are the circumstances when sealing and other protective mechanisms are likely to be effective, when they are likely to fail, and the zone of uncertainty between these two extremes.