Characterisation of equivalent permeability in heterogeneous fluvial architectural elements for CO2 storage: Bunter Sandstone Formation, UK

• Constructed architectural element scale models of braided fluvial sandstones using outcrop and modern analogues. • Evaluated equivalent permeability in three fluvial architectural elements for varying model volumes. • Performed facies clustering analysis to reduce model complexity while preserving flow-relevant heterogeneity. • Showed that even a simplified two-facies scheme preserves the key permeability architecture in most cases. • Provided practical guidance for simplifying fluvial reservoir models for CO 2 storage. Accurate characterisation of the effects of geological heterogeneity is critical for predicting fluid flow and storage in the subsurface. In reservoir modelling, capturing heterogeneity in detail and at multiple scales is required for such characterisation, but is rarely performed because it can be computationally intensive and time-consuming. This study provides the first quantitative assessment of the extent to which heterogeneity at the architectural-element scale in the Bunter Sandstone Formation can be simplified without significantly compromising estimation of equivalent permeability. For this, previously established core facies associations at the Endurance CO 2 storage site were correlated with outcrop facies associations, and based on their stacking patterns and geometry in outcrop analogues, three types of architectural elements (channel fills, lateral- and downstream-accreting bars, overbank deposits) were identified. Architectural-element models were then constructed using a sketch-based approach, with dimensions and geometries informed by outcrop and modern analogues. Three reservoir-modelling scenarios were evaluated: a baseline model incorporating all the constituent lithofacies, an intermediate model grouping lithofacies into three permeability-based categories, and a simplified two-facies model distinguishing between reservoir and non-reservoir categories. Equivalent permeability at varying model volumes was assessed along depositional dip, strike, and vertical directions for each scenario. The three-facies model showed permeability differences within 10–15% of the baseline model across most directions and model volumes. Although the two-facies model showed higher variability, especially in permeability parallel to depositional dip, permeability variations are still within 20% of the baseline model in most full-scale models. The largest deviations occurred in smaller models due to limited representation of the lithofacies variability. These results demonstrate that architectural-element-scale models can be effectively simplified using a two-facies approach for many applications, providing a practical balance between computational efficiency and model accuracy.