Capillary heterogeneity upscaling using macroscopic percolation: code advances and field-scale dynamic CO2 storage simulations

Our previous studies revealed that the impact of small-scale capillary heterogeneities is crucial for accurately predicting carbon dioxide (CO₂) plume migration in geological CO₂ storage sites, such as the Endurance site in the UK.  While high-fidelity dynamic modelling would require excessive computational resources, conventional upscaling methods of geological models often result in dynamic models underestimating lateral CO₂ migrations.

A novel capillary-limit steady-state upscaling approach is based on macroscopic invasion percolation and addresses this accuracy-feasibility trade-off. It incorporates small-scale capillary effects into upscaled local water/gas saturation functions: capillary pressure and phase permeabilities. We are developing an open-source algorithm implementation to encourage industrial adoption of the approach [1]. We present the latest advances in the library's features and performance and numerical experiments on our public set of dynamic models of real CO₂ storage sites in the North Sea [2]. 

The latest library advances include substantial parallelisation, single-core optimisations, an optional hydrostatic term, support for anisotropic fine-scale permeability, a stochastic re-upscaling approach for porosity and permeability fields upscaled by averaging, library infrastructure, and more. Numerical experiments aim to assess the impact of upscaling under uncertainties in rock and multiphase flow properties. We also attempt to downscale CO₂ plumes simulated at coarse scales using data from the algorithm's percolation step, providing an estimate for fine-scale dynamics.

[1] https://github.com/ImperialCollegeLondon/StrataTrapper
[2] https://github.com/ImperialCollegeLondon/StrataTrapper-models