Quantifying the Uncertainty of CO2 Leakage in Geological Carbon Storage

One of society's main concerns about Geological Carbon Storage (GCS) is the leakage of CO2 that could eventually contaminate freshwater resources or reach the surface. Many stakeholders believe that a prediction of leakage probability below one percent is adequate to ensure the viability of a CO2 storage project. However, the problem is that to be certain in predicting a rare leakage event, an enormous amount of possible models and flow simulations are needed, which is translated into extensive and inefficient computing time.

This study aimed to analyze how the geological structures, particularly the reservoir top surface geometry, impact the flow and eventual leakage of CO2. We used the MRST - CO2lab software from SINTEF to perform Monte Carlo flow simulations on geological models with uncertain top surface. To estimate this leakage probability with confidence, we then used an importance sampling method to produce a generating function to build more models with top surfaces with higher leakage probability. Then we computed a correction by weighing the samples. This methodology drastically reduces the number of models, flow simulations, and computing power to get to the desired confidence interval of the leakage probability. It shows a way to reduce the uncertainty of the different risk factors associated with GCS, thus allowing the planning of storage locations needed for viable GCS projects. Furthermore, this equips society and policymakers with guarantees to rapidly deploy this relevant carbon-neutral technology on a big scale.