Abstract
We use simplified methods to investigate how uncertainty in geological models affects practical CO2 storage capacities in large-scale saline aquifers. Our focus is on uncertainties in top-surface elevation, rock properties (porosity, permeability), fault transmissibility, and aquifer conditions (pressure and temperature). To quantify the statistical characteristics of static trapping capacity and dynamic estimates of plume migration, we create hundreds of possible realizations of the geomodel by applying Gaussian-type perturbations to the spatially correlated properties. Two different simplified methods are introduced to reduce the computational cost of simulating migration over thousands of years in all the model realizations, which each spans hundreds of kilometers. First, we use vertical-equilibrium (VE) modelling, which is orders of magnitude faster than solving the 3D flow equations. Second, we introduce a fast look-ahead algorithm that enables us to exit the VE simulation once a pseudo-steady state is reached. This algorithm uses a spill-point analysis of the top-surface's trapping structure to forecast how much CO2 will eventually become trapped and how much will leak through open boundaries of the formation. This reduces the computational cost significantly, since we seldom need to simulate long-term migration past a few hundred or thousand years.
