InterPore2026

A Statistical Approach to Determine REVs for Porosity and Permeability in Vesicular Basalts

20 May 2026, 10:05

1h 30m

Poster Presentation (MS09) Pore-Scale Physics and Modeling Poster

Speaker

Foojan Kazemzadeh Haghighi (School of Geography, Earth and Atmospheric Sciences, University of Melbourne)

Description

Introduction

The determination of the representative elementary volume (REV) is fundamental for predicting large-scale fluid flow behavior in porous rocks (Singh et al., 2020). While REV analysis is well-established for sedimentary rocks like sandstones and limestones, the complex and highly heterogeneous pore architecture of vesicular basalts remains poorly studied. Addressing this knowledge gap is particularly important given the global interest in CO$_2$ geo-sequestration, where basaltic formations are targeted due to their capacity to rapidly mineralize CO$_2$ (Metz et al., 2005; Snæbjörnsdóttir et al., 2020).

Methodology
This paper investigates the statistical REV (sREV) of vesicular basalt samples from Port Fairy, Australia, focusing on porosity and permeability. The methodology utilized denoised and processed 3D micro-CT images (Fig. 1). In order to overcome the very large and impractical computational demands of direct numerical simulations (DNS) when computing permeability, a graph-based approach was employed. The reduced physics proxy is referred to as the least resistance index (LRI), where voxels are treated as nodes to identify the path of minimum resistance via a cost function (Mishra et al., 2024). The inverse relationship between LRI and permeability results was validated against DNS using the GeoChemFoam code (Menke et al., 2021). Following petrophysical extraction, three statistical methods determined the sREV onset. Beyond the standard coefficient of variation (CV) trend, we implemented a robust parameter sweep analysis. This second approach evaluated convergence by applying thresholds to CV derivative ratios and requiring these criteria to be satisfied across windows of sub-sample points. The final method for REV onset used bootstrap resampling to generate confidence intervals and an overall score for stability.

Results & Discussions
The investigation revealed that vesicular basalts exhibited extreme heterogeneity, requiring sub-volumes over 1000 times larger than those typically required for sandstones and carbonates in order to reach statistical representativeness (Fig. 2). While the Bentheimer Sandstone benchmark achieved permeability and porosity REV at volumes between 1 mm$^3$ – 14 mm$^3$, the cubic basalt sub-samples failed to converge even when considering a sub-volume of 13,800 mm$^3$. In contrast, the REV in cuboidal sub-samples was identified within a lower and upper bound of 18,000 mm$^3$ and 43,000 mm$^3$. Furthermore, LRIs were analyzed in three principal directions, which showed that the REV is highly dependent on sample orientation due to anisotropy in pore connectivity. Permeability validation showed a close match between experimental (109±20 mD) and numerically simulated (123±30 mD) values, confirming the accuracy of the image processing steps as well as the dominance of the interconnected vesicular network on the bulk flow. Additionally, two-point correlation analysis linked the large REV requirements to the extended correlation lengths and structural diversity of the basaltic pore space. Finally, analysis of a larger sample revealed localized regions of zero effective porosity, contrasting with the connected networks identified in the other samples. This disparity emphasized that pore connectivity in vesicular basalts is highly non-uniform, suggesting that different sections of the vesicular layer may possess a distinct and unique REV.