Probabilistic CO₂ plume modelling

Abstract

The Offshore Petroleum and Greenhouse Gas Storage Act 2006 provides the legal framework for carbon dioxide (CO₂) storage projects in Australian Commonwealth waters, with the associated regulations and guidelines requiring that applicants for a ‘declaration of an identified storage formation’ consider ‘all migration pathways of which the probability of occurrence is greater than 10%’. The overseas Society of Petroleum Engineer’s CO₂ Storage Resources Management System goes further and discusses the concept of modelling not only the subsurface CO₂ plume itself but also the attendant pressure effects and displaced formation water, which, in that scheme, comprise part of a wider ‘containment area’. These collective requirements necessitate that the eligible carbon storage resources be assessed probabilistically, a requirement that can result in sizable static and dynamic models that are computationally heavy to simulate. This computational complexity places inherent, practical limitations on using ‘conventional’ probabilistic workflows, such as Monte Carlo simulation or Latin hypercube sampling. To address this limitation, factorial experimental designs were employed in the modelling workflow. Factorial designs are widely used in settings where individual experiments are costly or require prolonged periods and, when implemented correctly, can uncover underlying uncertainty distributions using a minimum number of simulation cases. The efficacy and power of this approach is demonstrated using CO₂ injection and migration models for a carbon storage project located in offshore Western Australia, which is in the early development stage. The ability to test potential sensitivities rapidly has proven to be critical in refining and defining the project’s ultimate carbon capture and storage development plan.