|Abstract:|| Modelling the highly localised plume spreading during CO2 geo-sequestration using conventional synchronous time-driven simulation (TDS) has been impeded by the stringent Courant-Fredrich-Levy (CFL) condition, which leads to an excessive number of time steps and consequently long computing times. To overcome this problem, we present an asynchronous discrete event simulation (DES) scheme based on local time stepping criteria, specifically developed for the CSMP++ CO2 geo-sequestration simulator. The proposed DES method is applied to a complex and heterogeneous heuristic CO2 storage model, where it proves that DES is able to concentrate the computational effort on the active regions where fast CO2 flow occurs. As a result, the execution time for the modelling of a 5-year injection is significantly reduced from over 91 days (estimated for TDS) to only 0.5 days. This dramatic speedup facilitates the modelling of CO2 injection and long-term plume spreading behaviours at the scales of field storage sites. The benefits of the new method scale with the level of refinement of geologic detail and include a distinct increase in the level of physical realism of the simulations because fast and slow events are equally well resolved in contrast with TDS implicit schemes which are robust, but fail to resolve the events captured by the new asynchronous scheme.