Uncertainties in practical simulation of CO2 storage

Practical simulation of CO2 storage in geological formations inherently involves decisions concerning relevant physics, upscaling, and numerical modeling. These decisions are unavoidable, since the full problem cannot be resolved by existing numerical approaches. Here, we report on the impact of three distinct approaches to make the problem computationally tractable: reduced physics, upscaling, and non-converged discretizations. Compounding these different strategies, we have used a benchmark study to try to assess the impact of an expert group on the results of the numerical simulations. In order to restrict the scope of the investigation, the geometric and geological description of the storage aquifer was simplified to the greatest extent possible.The different strategies applied to simplify the problem, lead to significantly deviating answers when addressing relevant storage questions. Furthermore, there is room for interpretation when complex simulation results are simplified to the type of higher-level information sought in decision making processes. Our experience leads us to conclude that, important questions relating to CO2 storage cannot be predicted convincingly to satisfactory accuracy with numerical simulation tools, even for highly idealized problems. This emphasizes the need for real-time monitoring and history matching during injection operations.

► Benchmark study focusing on the coupled physics of CO2 storage.
► Relevant and important questions for CO2 storage cannot be predicted accurately.
► Uncertainty in parameters cannot alone explain the variability of reported results.
► Ensemble uncertainty is much greater than any individual sensitivity study.
► Numerical models should not be mistaken for accurate prognostic tools.