Assessing the impact of sedimentary heterogeneity on CO2 injection in fluvial meander-belt successions using geostatistical modelling informed by geological analogues

Abstract

To optimize programmes of carbon capture and storage, it is crucial to understand how subsurface heterogeneity may control CO₂ dispersal in sedimentary reservoir successions. It is therefore necessary to evaluate the impact of subsurface modelling techniques on predictions in lithological and petrophysical heterogeneity, and on resulting dynamic behaviours. In this study, alternative idealized, unconditional static models were created that incorporate different types of sedimentary heterogeneities typical of fluvial meander-belt sedimentary successions, at different scales. These static models were produced using two different geostatistical algorithms based on multipoint statistics: SNESIM and DEESSE. Two alternative sets of geocellular grids were created that capture (i) macroscale levels of heterogeneity only (architectural elements) and (ii) both macro- and mesoscale (point-bar lithologies) heterogeneities, respectively. The geocellular models were populated with petrophysical data from a selected geological analogue (Barracouta Formation, Australia), imposing a depth-related trend based on the analysis of literature data. Porosity and permeability models were obtained via Gaussian random function simulations. These static models were used to simulate subsurface CO₂ injection over a 30-year period to enable tracking of plume propagation and a comparison between models incorporating different levels of facies heterogeneity. The study highlights the influence of the underlying facies framework on CO₂ dynamic simulations, since aspects of reservoir pressure redistribution and caprock pressure relief only emerge from models incorporating mesoscale features. Furthermore, predicted CO₂ plume displacement, injection rates and cumulative injected volumes are also affected by the facies-modelling approach. Modelling categories and strategies must be carefully selected in subsurface modelling workflows applied to plan CCS projects.