摘要
In this paper, we develop a methodology for early detection of potential CO2 leakage from geological storage formations using pressure and surface-deformation anomalies. The basic idea is based on the fact that leakage-induced pressure signals travel much faster than the migrating CO2; thus such anomalies may be detected early enough for risk management measures taking effect in avoiding substantial CO2 leaks. The early detection methodology involves automatic inversion of anomalous brine leakage signals with efficient forward pressure and surface-deformation modeling tools to estimate the location and permeability of leaky features in the caprock. We conduct a global sensitivity analysis to better understand under which conditions pressure anomalies can be clearly identified as leakage signals, and evaluate signal detectability for a broad parameter range considering different detection limits and levels of data noise. The inverse methodology is then applied to two synthetic examples of idealized two-aquifer-and-one aquitard storage systems, with an injection well and a leaky well, for different monitoring scenarios. In Example 1, only pressure data at the monitoring and injection wells are used for leakage detection. Our results show that the accuracy of leakage detection greatly depends on the level of pressure data noise. In Example 2, joint inversion of pressure and surface-deformation measurements significantly improves the speed of convergence toward the true solution of the leakage parameters and enables early leakage detection. In both examples, successful detection is achieved when two monitoring wells are appropriately placed within up to 4 km from the leaky well.